Publications

Authoring a multi-actor scenario for a living 3D world, where every action changes its state, and each action’s validity depends on the state accumulated before it, demands the freedom of storytelling and the rigor of simulation at once. We author such scenarios with LLM agents, as Graphs of Events in Space and Time (GESTs) that a simulation engine executes deterministically into narrative videos with per-frame spatial, temporal, and semantic ground truth. A staged pipeline driving a flagship LLM, the standard design in video generation, failed outright: the model violates rules stated verbatim in its prompt, and cannot track the dynamic world state. We answer with a constraint-enforcing tool layer: our Director and Scene Builder agents explore the world’s capabilities page by page and build every scene through operations checked against simulator state, so every specification they emit is valid by construction. Because we generate each seed text from an existing scenario graph, we can measure reconstruction: the agent authors its own graph from the text alone, yet matches the original at 0.83 F1 on its events, each action with its participants (0.55 for a random scenario of the same kind), and 0.77 on their ordering (0.43 random). End to end: the standard staged pipeline produced 0 executable specifications in 50 attempts; our agents, driving a budget model, execute 20 of 25 (80%), and are, to our knowledge, the first to exercise the full expressive capacity of GEST.

Current methods for detecting spurious correlations rely on analyzing dataset statistics or error patterns, leaving many harmful shortcuts invisible when counterexamples are absent. We introduce BEE (Bridging Explainability and Embeddings), a framework that shifts the focus from model predictions to the weight space, and to the embedding geometry underlying decisions. By analyzing how fine-tuning perturbs pretrained representations, BEE uncovers spurious correlations that remain hidden from conventional evaluation pipelines. We use linear probing as a transparent diagnostic lens, revealing spurious features that not only persist after full fine-tuning but also transfer across diverse state-of-the-art models. Our experiments cover numerous datasets and domains: vision (Waterbirds, CelebA, ImageNet-1k), language (CivilComments, MIMIC-CXR medical notes), and multiple embedding families (CLIP, CLIP-DataComp.XL, mGTE, BLIP2, SigLIP2). BEE consistently exposes spurious correlations: from concepts that slash the ImageNet accuracy by up to 95%, to clinical shortcuts in MIMIC-CXR notes that induce dangerous false negatives. Together, these results position BEE as a general and principled tool for diagnosing spurious correlations in weight space, enabling principled dataset auditing and more trustworthy foundation models. The source code is publicly available at https://github.com/bit-ml/bee.

While Deep Learning has demonstrated impressive results in applications on various data types, it continues to lag behind tree-based methods when applied to tabular data, often referred to as the last “unconquered castle” for neural networks. We hypothesize that a significant advantage of tree-based methods lies in their intrinsic capability to model and exploit non-linear interactions induced by features with categorical characteristics. In contrast, neural-based methods exhibit biases toward uniform numerical processing of features and smooth solutions, making it challenging for them to effectively leverage such patterns. We address this performance gap by using statistical-based feature processing techniques to identify features that are strongly correlated with the target once discretized. We further mitigate the bias of deep models for overly-smooth solutions, a bias that does not align with the inherent properties of the data, using Learned Fourier. We show that our proposed feature preprocessing significantly boosts the performance of deep learning models and enables them to achieve a performance that closely matches or surpasses XGBoost on a comprehensive tabular data benchmark.

We present an efficient and robust method for 3D geometric reconstruction that is based solely on the camera-independent linear relationships among a given set of points, which are stable over time and robustly estimated using multiple point matches. We essentially learn, from correspondences between points across several frames, a linear geometric auto-regression matrix $\mathbf{W}$, which establishes how a point in 3D can be expressed as a linear combination of all the others. This matrix is constant and does not depend on the world coordinate system or the camera pose—it is an intrinsic property of the point set. We also show that the principal eigenvectors of $\mathbf{W}$, which all have eigenvalue $1$, provide a homogeneous representation of the 3D point configuration. The first version of our method takes advantage of noisy monocular depth maps in order to obtain, from multiple frames, a robust geometric auto-regression matrix $\mathbf{W}$ of linear relationships between the 3D points. Thus, we build on recent advances in deep learning, which now provide monocular depth estimation models that are fast but very often noisy. Our approach handles noise through robust linear estimation over several frames. The second version of our method does not need monocular depth estimation maps. It applies in cases of weak-perspective projection, when the linear combinations between the 3D points can be robustly estimated from their 2D projections in the image. Note that the camera projection matrix is never used in our derivations. Consequently, our method does not recover camera pose, but only 3D structure. This is a key difference between our method and the related literature on 3D geometric reconstruction.

Game engines hold what video models struggle to learn: a complete, explicit world state behind every frame. We turn one into a data instrument. GEST-Engine, our production-grade open-source system, deterministically executes Graphs of Events in Space and Time (GESTs), whether procedurally generated or derived from text, into videos of synchronized multi-actor scenarios, recording ground truth as it renders: 3D entity and camera state, pairwise spatial relations, event-to-frame mappings, instance segmentation, and long descriptions, at zero marginal annotation cost. With it we release GTASA, a 938-video sample of what the system can generate at arbitrary scale, carrying, to our knowledge, the densest spatial-relation coverage of any video dataset: a complete entity-pair relation graph at every frame, ~84x denser than the state of the art, frame-for-frame. We validate GTASA both qualitatively, through human evaluation of physical validity and semantic alignment where frontier neural generators, given the same prompts, largely fail, and quantitatively, with GTASA pretraining improving VideoLLaMA3 video captioning. Probing six frozen video encoders across 11 spatio-temporal tasks enabled by GTASA’s exact 3D ground truth, a previously untestable inter-entity relational probe of frozen video features, reveals that who-is-near-whom barely rises above chance for all of them. We release the engine, the corpus, and the benchmark, making this gap a measurable, trainable target.

Can a dataset be recognized from the spurious correlations it induces during training? We argue that datasets leave dataset-specific traces in a model’s learned semantic correlation structure: incidental regularities that are predictive within a dataset, but not causal for the underlying task, can be internalized during training. We use this insight to study dataset-level membership inference, moving beyond existing methods that rely on behavioral or distributional evidence such as confidence scores, losses, margins, generated samples, or query responses. We introduce a white-box semantic fingerprinting approach based on semantic correlation descriptors (SCDs), which capture the semantic correlation structure learned by a model and make it comparable across dataset mixtures. In a controlled leave-one-dataset-out diagnostic, SCDs recover dataset-specific changes and perfectly separate matching from non-matching dataset pairs. We then propose a practical SCD-based membership score that tests whether a target dataset is part of a model’s training mixture using only the model’s SCD and the target dataset’s standalone SCD, without requiring leave-one-dataset-out models. Across three diverse experimental settings, with dataset groups for natural language inference, emotion classification, and medical text classification, we test both the advantages and limitations of SCD-based membership inference with different degrees of semantic separation and keyword support between dataset splits. On average, the classifier based on this score achieves the highest performance and the lowest std, outperforming black-box baselines RMIA, Attack-P, and LiRA, as well as the white-box SIF baseline. These results show that dataset membership can be traced through internal semantic correlations, with the largest relative gain exceeding 60% in ROC-AUC when dataset groups expose distinct semantic particularities.

Vision-Language Models (VLMs) largely follow the text-only LLM trajectory, excelling on English benchmarks but sharply degrading on low-resource languages, where neither large-scale image-text corpora nor culturally grounded evaluations exist. We present a systematic study of building a language-specific VLM for Romanian, covering the full pipeline from data construction to architectural choices. We translate established English VLM training and evaluation corpora into Romanian, applying machine translation to textual annotations and to in-image text, preserving visual grounding while adapting the textual content. Using this data, we train and ablate a series of VLMs to isolate the contribution of (i) vision backbones of varying scale and pretraining, (ii) language backbones from multilingual to Romanian-adapted LLMs, and (iii) OCR-style image-text data. We further curate HoraVQA, a culturally native evaluation set grounded in Romanian everyday scenes. Romanian-adapted VLMs consistently outperform their same-sized counterparts and, across all evaluated benchmarks, even surpass models from the next larger size category.

Our team, VANGUARD, presents IROH (Insightful Ranking of Humor), a three-stage retrieval system for JOKER Task 1 English at CLEF 2026, achieving first place on the leaderboard with 0.6347 MAP. Our pipeline combines hybrid sparse-dense retrieval, cross-encoder reranking, and a LoRA-adapted Large Language Model judge ensemble. We employ Gemma 4 to generate query-aware rationales under two prompt strategies, generic and typed, and produce up to four types of structured hard negatives for training data construction. Through an ablation across three cross-encoder architectures, four dense embedders, and eight judge configurations, our key findings are threefold: (1) the rationale-distilled judge is the primary driver of ranking quality, whereas appending rationales to the first-stage index contributes negligibly; (2) structured hard negatives degrade generalisation in nearly all configurations despite inflating local validation scores; and (3) across the components we ablate, the lighter, better-calibrated model is competitive with or stronger than its larger counterpart, with the generic-rationale Qwen2.5-7B judge (0.6055 MAP) outperforming every Gemma-4-31B configuration, and the advantage of generic over typed rationales is concentrated almost entirely in the smaller model.

Adapter-based methods have become a cost-effective approach to continual learning (CL) for Large Language Models (LLMs), by sequentially learning a low-rank update matrix for each task. To mitigate catastrophic forgetting, state-of-the-art approaches impose constraints on new adapters with respect to the previous ones, by targeting either subspace or coordinate-wise interference. In this paper, we propose JumpLoRA, a novel framework to adaptively induce sparsity in the Low-Rank Adaptation (LoRA) blocks through the use of JumpReLU gating. The method achieves dynamic parameter isolation, which helps prevent task interference. We demonstrate that our method is highly modular and compatible with LoRA-based CL approaches. Specifically, it significantly boosts the performance of IncLoRA and outperforms the leading state-of-the-art CL method, ELLA.

Autonomous agents such as indoor drones must learn new object classes in real-time while limiting catastrophic forgetting, motivating Class-Incremental Learning (CIL). However, most unmanned aerial vehicle (UAV) datasets focus on outdoor scenes and offer limited temporally coherent indoor videos. We introduce an indoor dataset of $14,400$ frames capturing inter-drone and ground vehicle footage, annotated via a semi-automatic workflow with a $98.6%$ first-pass labeling agreement before final manual verification. Using this dataset, we benchmark 3 replay-based CIL strategies: Experience Replay (ER), Maximally Interfered Retrieval (MIR), and Forgetting-Aware Replay (FAR), using YOLOv11-nano as a resource-efficient detector for deployment-constrained UAV platforms. Under tight memory budgets ($5-10%$ replay), FAR performs better than the rest, achieving an average accuracy (ACC, $mAP_{50-95}$ across increments) of $82.96%$ with $5%$ replay. Gradient-weighted class activation mapping (Grad-CAM) analysis shows attention shifts across classes in mixed scenes, which is associated with reduced localization quality for drones. The experiments further demonstrate that replay-based continual learning can be effectively applied to edge aerial systems. Overall, this work contributes an indoor UAV video dataset with preserved temporal coherence and an evaluation of replay-based CIL under limited replay budgets. Project page: https://spacetime-vision-robotics-laboratory.github.io/learning-on-the-fly-cl

Perceiving human motion and intent at long range is a prerequisite for socially intelligent aerial robots, yet the data to learn it barely exists. We introduce Drones2BodyLanguage, a dataset grounding human motion in real UAV footage: avatars manifesting ten communicative intents are placed into unmodified 4K drone scenes with metrically correct position, scale and orientation, maintained over hundreds of frames of camera motion. Enabling it is a lightweight geometric world model of the local scene - semantically selected anchors lifted to 3D through streaming monocular depth - in which a placement point is predicted as an affine anchor combination with provably rigid-invariant weights, and re-rendered under an SVD-fitted ground rotation. Across twelve architectures on scene- and motion-disjoint splits, training on placed data lifts mean intent accuracy by a wide margin for real, retargeted and generated motion alike, with gains confirmed on two in-the-wild scenes.

Body movement communicates intent at distances and in conditions where neither the face, nor speech can be captured. We study the recognition of communicative intent from 2D body pose alone. We argue that body motion is a reliable signal especially in scenarios that require real time low-cost on-device person-to-robot communication in long distance environments, such as rescue missions. However, existing resources do not isolate this signal. Affective corpora combine body, face, voice and text, while skeleton action-recognition benchmarks label the action performed rather than the message conveyed. We release a dataset of real frames of full-body pose covering ten communicative intents and we compare it against other real (IPC) and synthetic (MotionLCM, VEO3.1, Kimodo) ones that span a range of difficulty. We target systems that can run on a robot’s limited onboard hardware. We benchmark multiple models, from skeleton graph classifiers to joint motion-forecasting networks, and report performance metrics together with frame rate on an embedded GPU (NVIDIA Orin~Nano), since speed matters as much as accuracy in our scenario. Finally, we show that a model’s own autoregressive self-consistency works as an unsupervised reliability signal. We give a short proof that bounds the probability that a self-consistent prediction is correct, show that this probability grows with the number of consistent steps, and identify the conditions under which a confident prediction can still be false, benchmarked against industry-standard metrics.

We study the ongoing debate regarding the statistical fidelity of AI-generated data compared to human-generated data in the context of non-verbal communication using full body motion. Concretely, we ask if contemporary generative models move beyond surface mimicry to participate in the silent, but expressive dialogue of body language. We tackle this question by introducing the first framework that generates a natural non-verbal interaction between Human and AI in real-time from 2D body keypoints. Our experiments utilize four lightweight architectures which run at up to 100 FPS on an NVIDIA Orin Nano, effectively closing the perception-action loop needed for natural Human-AI interaction. We trained on 437 human video clips and demonstrated that pretraining on synthetically-generated sequences reduces motion errors significantly, without sacrificing speed. Yet, a measurable reality gap persists. When the best model is evaluated on keypoints extracted from cutting-edge text-to-video systems, such as SORA and VEO, we observe that performance drops on SORA-generated clips. However, it degrades far less on VEO, suggesting that temporal coherence, not image fidelity, drives real-world performance. Our results demonstrate that statistically distinguishable differences persist between Human and AI motion.

The computer vision domain has greatly benefited from an abundance of data across many modalities to improve on various visual tasks. Recently, there has been a lot of focus on self-supervised pre-training methods through Masked Autoencoders (MAE) \cite{he2022masked,bachmann2022multimae}, usually used as a first step before optimizing for a downstream task, such as classification or regression. This is very useful as it doesn’t require any manually labeled data. In this work, we introduce Probabilistic Hyper-Graphs using Masked Autoencoders (PHG-MAE): a novel model that unifies the classical work on neural graphs \cite{leordeanu2021semi} with the modern approach of masked autoencoders under a common theoretical framework. Through random masking of entire modalities, not just patches, the model samples from the distribution of hyper-edges on each forward pass. Additionally, the model adapts the standard MAE algorithm by combining pre-training and fine-tuning into a single training loop. Moreover, our approach enables the creation of inference-time ensembles which, through aggregation, boost the final prediction performance and consistency. Lastly, we show that we can apply knowledge distillation on top of the ensembles with little loss in performance, even with models that have fewer than 1M parameters. While our work mostly focuses on outdoor UAV scenes that contain multiple world interpretations and modalities, the same steps can be followed in other similar domains, such as autonomous driving or indoor robotics. In order to streamline the process of integrating external pre-trained experts for computer vision multi-modal multi-task learning (MTL) scenarios, we developed a data-pipeline software. Using this tool, we have created and released a fully-automated extension of the Dronescapes dataset. All the technical details, code and reproduction steps are publicly released.

Continual learning (CL) examines how models learn a sequence of tasks while retaining previously learned knowledge. Despite substantial progress in benchmarking CL methods, comparative evaluations typically keep the fine-tuning regime fixed. In this paper, we argue that the fine-tuning regime, defined by the trainable parameter subspace, is itself a key evaluation variable. We formalize adaptation regimes as projected optimization over fixed trainable subspaces, showing that changing the trainable depth alters the effective update signal through which both current task fitting and knowledge preservation operate. This analysis motivates the hypothesis that method comparisons need not be invariant across regimes. We test this hypothesis in task incremental CL while considering 5 trainable depth regimes and 5 standard methods: online EWC, LwF, SI, GEM, and DER. We find that the relative ranking of methods is not consistently preserved across regimes when evaluating across 5 benchmark datasets, namely MNIST, Fashion MNIST, KMNIST, QMNIST, and CIFAR-100, and across 11 task orders per dataset. We further show that deeper adaptation regimes are associated with larger update magnitudes, higher forgetting, and a stronger relationship between the two. These results show that comparative conclusions in CL can depend strongly on the chosen fine-tuning regime, motivating regime-aware evaluation protocols that treat trainable depth as an explicit experimental factor.

Streaming Continual Learning (CL) typically converts a continuous stream into a sequence of discrete tasks through temporal partitioning. We argue that this temporal taskification step is not a neutral preprocessing choice, but a structural component of evaluation: different valid splits of the same stream can induce different CL regimes and therefore different benchmark conclusions. To study this effect, we introduce a taskification-level framework based on plasticity and stability profiles, a profile distance between taskifications, and Boundary-Profile Sensitivity (BPS), which diagnoses how strongly small boundary perturbations alter the induced regime before any CL model is trained. We evaluate continual finetuning, Experience Replay, Elastic Weight Consolidation, and Learning without Forgetting on network traffic forecasting with CESNET-Timeseries24, keeping the stream, model, and training budget fixed while varying only the temporal taskification. Across 9-, 30-, and 44-day splits, we observe substantial changes in forecasting error, forgetting, and backward transfer, showing that taskification alone can materially affect CL evaluation. We further find that shorter taskifications induce noisier distribution-level patterns, larger structural distances, and higher BPS, indicating greater sensitivity to boundary perturbations. These results show that benchmark conclusions in streaming CL depend not only on the learner and the data stream, but also on how that stream is taskified, motivating temporal taskification as a first-class evaluation variable.

We present the GEST-Engine, a complete system that goes from natural-language text to fully-annotated multi-actor video. At its core is an explicit world model: rather than encoding state as a learned latent, the engine maintains a complete, inspectable representation of the world (which actors exist, where they are, what they are doing, which objects they hold, and how events relate in time and space), expressed as a formal Graph of Events in Space and Time (GEST) and realized deterministically inside the open world of a commercial game engine driven through an open-source multiplayer scripting framework. GESTs are produced either procedurally or by an agentic text-to-GEST system in which an LLM Director plans a story through tool calls validated by a programmatic state backend, so every generated specification is executable by construction. A GEST then enters a four-stage execution pipeline: graph parsing and validation, entity and action grounding, temporal orchestration (Allen-style constraints resolved by Floyd-Warshall transitive closure), and execution and capture. In a single simulation pass the engine emits frame-aligned RGB video, dense per-pixel depth, instance segmentation, per-actor skeletal pose, per-frame pairwise spatial-relation graphs, 2D bounding boxes, event-to-frame temporal mappings, and natural-language descriptions, all at zero marginal annotation cost. We further describe an in-game world editor, runtime capability extraction, a text-generation pipeline, and a production system that renders corpora at scale across parallel virtual machines. Because every frame traces back to a semantic specification, the engine guarantees object permanence, multi-actor coordination, and temporal consistency by construction, making its output valuable as training data, evaluation benchmarks, and diagnostic tools for video understanding.

Detecting sexism on the internet is a fundamentally subjective task; our team, VANGUARD, addresses this challenge in the EXIST 2026 Task 2 by proposing a human-centered multimodal framework that analyses and incorporates the psychological and demographic characteristics of human annotators into the detection pipeline. We fuse five input modalities through a cross-attention architecture with Feature-wise Linear Modulation conditioning. Meme text is extracted and visually described with Gemma 4, then augmented by automatic translation between English and Spanish with NLLB-200. Text and image representations are produced by LoRAadapted XLM-RoBERTa and CLIP encoders and fused with sensor features encoded by a pretrained autoencoder. To model annotator subjectivity, we frame Subtask 2.1 as a label distribution learning problem, optimizing a Kullback-Leibler divergence loss over the full annotator label distribution. At inference time, predictions are produced by soft-voting between the deep multimodal network and a complementary SVM trained on stylometric and physiological features. Our best submission ranks 29th out of 114 on Subtask 2.2 (source intention) under soft evaluation, and the normalized ICM scores remain above the baseline on Subtasks 2.1 and 2.2, indicating that annotator-centered conditioning contributes a usable signal. We release our full pipeline and analysis to support reproducible human-centered modeling.

We present ChronoGraph, a graph-structured multivariate time series forecasting dataset built from real-world production microservices. Each node is a service that emits a multivariate stream of system-level performance metrics, capturing CPU, memory, and network usage patterns, while directed edges encode dependencies between services. The primary task is forecasting future values of these signals at the service level. In addition, ChronoGraph provides expert-annotated incident windows as anomaly labels, enabling evaluation of anomaly detection methods and assessment of forecast robustness during operational disruptions. Compared to existing benchmarks from industrial control systems or traffic and air-quality domains, ChronoGraph uniquely combines (i) multivariate time series, (ii) an explicit, machine-readable dependency graph, and (iii) anomaly labels aligned with real incidents. We report baseline results spanning forecasting models, pretrained time-series foundation models, and standard anomaly detectors. ChronoGraph offers a realistic benchmark for studying structure-aware forecasting and incident-aware evaluation in microservice systems.

This work introduces a self-supervised neuro-analytical, cost efficient, model for visual-based quadrotor control in which a small 1.7M parameters student ConvNet learns automatically from an analytical teacher, an improved image-based visual servoing (IBVS) controller. Our IBVS system solves numerical instabilities by reducing the classical visual servoing equations and enabling efficient stable image feature detection. Through knowledge distillation, the student model achieves 11x faster inference compared to the teacher IBVS pipeline, while demonstrating similar control accuracy at a significantly lower computational and memory cost. Our vision-only self-supervised neuro-analytic control, enables quadrotor orientation and movement without requiring explicit geometric models or fiducial markers. The proposed methodology leverages simulation-to-reality transfer learning and is validated on a small drone platform in GPS-denied indoor environments. Our key contributions include: (1) an analytical IBVS teacher that solves numerical instabilities inherent in classical approaches, (2) a two-stage segmentation pipeline combining YOLOv11 with a U-Net-based mask splitter for robust anterior-posterior vehicle segmentation to correctly estimate the orientation of the target, and (3) an efficient knowledge distillation dual-path system, which transfers geometric visual servoing capabilities from the analytical IBVS teacher to a compact and small student neural network that outperforms the teacher, while being suitable for real-time onboard deployment.

The task of describing video content in natural language is commonly referred to as video captioning. Unlike conventional video captions, which are typically brief and widely available, long-form paragraph descriptions in natural language are scarce. This limitation of current datasets is due to the expensive human manual annotation required and to the highly challenging task of explaining the language formation process from the perspective of the underlying story, as a complex system of interconnected events in space and time. Through a thorough analysis of recently published methods and available datasets, we identify a general lack of published resources dedicated to the problem of describing videos in complex language, beyond the level of descriptions in the form of enumerations of simple captions. Furthermore, while state-of-the-art methods produce impressive results on the task of generating shorter captions from videos by direct end-to-end learning between the videos and text, the problem of explaining the relationship between vision and language is still beyond our reach. In this work, we propose a shared representation between vision and language, based on graphs of events in space and time, which can be obtained in an explainable and analytical way, to integrate and connect multiple vision tasks to produce the final natural language description. Moreover, we also demonstrate how our automated and explainable video description generation process can function as a fully automatic teacher to effectively train direct, end-to-end neural student pathways, within a self-supervised neuro-analytical system. We validate that our explainable neuro-analytical approach generates coherent, rich and relevant textual descriptions on videos collected from multiple varied datasets, using both standard evaluation metrics, human annotations and consensus from ensembles of state-of-the-art VLMs.

Indoor navigation is a difficult task, as it generally comes with poor GPS access, forcing solutions to rely on other sources of information. While significant progress continues to be made in this area, deployment to production applications is still lacking, given the complexity and additional requirements of current solutions. Here, we introduce an efficient, real-time and easily deployable deep learning approach, based on visual input only, that can predict the direction towards a target from images captured by a mobile device. Our technical approach, based on a novel graph-based path generation method, combined with explainable data augmentation and curriculum learning, includes contributions that make the process of data collection, annotation and training, as automatic as possible, efficient and robust. On the practical side, we introduce a novel large-scale dataset, with video footage inside a relatively large shopping mall, in which each frame is annotated with the correct next direction towards different specific target destinations. Different from current methods, ours relies solely on vision, avoiding the need of special sensors, additional markers placed along the path, knowledge of the scene map or internet access. We also created an easy to use application for Android, which we plan to make publicly available. We make all our data and code available along with visual demos on our project site11https://sites.google.com/view/indoor-navigation-by-vision/.

Pulmonary Embolism (PE) are a leading cause of cardiovascular death. Computed tomographic pulmonary angiography (CTPA) is the gold standard for PE diagnosis, with growing interest in AI-based diagnostic assistance. However, these algorithms are limited by scarce fine-grained annotations of thromboembolic burden. We address this challenge with iExplain, a weakly supervised learning algorithm that transforms coarse image-level annotations into detailed pixel-level PE masks through iterative model explainability. Our approach generates soft segmentation maps used to mask detected regions, enabling the process to repeat and discover additional embolisms that would be missed in a single pass. This iterative refinement effectively captures complete PE regions and detects multiple distinct embolisms. Models trained on these automatically generated annotations achieve excellent PE detection performance, with significant improvements at each iteration. We demonstrate iExplain’s effectiveness on the RSPECT augmented dataset, achieving results comparable to strongly supervised methods while outperforming existing weakly supervised methods.

In order to evaluate the impact of k-means clustering on portfolio optimization, this study groups enterprises based on profitability, liquidity, and solvency indicators. The study confirms the positive correlation between risk, return, and risk-adjusted performance through an analysis of historical financial records. After the companies were divided into two groups, equal-weighted portfolios were created using these groupings. Although they produced higher returns, cluster 1 portfolios, which included more risky companies, also showed more volatility. Cluster 0 portfolios, on the other hand, offered less risk and more consistent results. Portfolios clustered by ROA, OCFM, and GPM outperformed the market benchmark and produced the highest returns adjusted for risk, according to Sharpe Ratio analysis. Furthermore, the study emphasizes that although solvency and liquidity metrics play a role in portfolio selection, increased liquidity does not always translate into improved risk-adjusted performance. In terms of methodology, Silhouette Analysis outperformed the Elbow technique in determining the optimal number of clusters. All things considered, the results show how data-driven clustering techniques may be used to align portfolio strategies to investors’ risk tolerances.

We pose video object segmentation as spectral graph clustering in space and time, with one graph node for each pixel and edges forming local space-time neighborhoods. We claim that the strongest cluster in this video graph represents the salient object. We start by introducing a novel and efficient method based on 3D filtering for approximating the spectral solution, as the principal eigenvector of the graph’s adjacency matrix, without explicitly building the matrix. This key property allows us to have a fast parallel implementation on GPU, orders of magnitude faster than classical approaches for computing the eigenvector. Our motivation for a spectral space-time clustering approach, unique in video semantic segmentation literature, is that such clustering is dedicated to preserving object consistency over time, which we evaluate using our novel segmentation consistency measure. Further on, we show how to efficiently learn the solution over multiple input feature channels. Finally, we extend the formulation of our approach beyond the segmentation task, into the realm of object tracking. In extensive experiments we show significant improvements over top methods, as well as over powerful ensembles that combine them, achieving state-of-the-art on multiple benchmarks, both for tracking and segmentation.

We introduce Generalized Test-Time Augmentation (GTTA), a highly effective method for improving the performance of a trained model, which unlike other existing Test-Time Augmentation approaches from the literature is general enough to be used off-the-shelf for many vision and non-vision tasks, such as classification, regression, image segmentation and object detection. By applying a new general data transformation, that randomly perturbs multiple times the PCA subspace projection of a test input, GTTA creates valid augmented samples from the data distribution with high diversity, properties we theoretically show that are essential for a Test-Time Augmentation method to be effective. Different from other existing methods, we also propose a final self-supervised learning stage in which the ensemble output, acting as an unsupervised teacher, is used to train the initial single student model, thus reducing significantly the test time computational cost. Our comparisons to strong TTA approaches and SoTA models on various vision and non-vision well-known datasets and tasks, such as image classification and segmentation, pneumonia detection, speech recognition and house price prediction, validate the generality of the proposed GTTA. Furthermore, we also prove its effectiveness on the more specific real-world task of salmon segmentation and detection in low-visibility underwater videos, for which we introduce DeepSalmon, the largest dataset of its kind in the literature.

The real-world is inherently multi-modal at its core. Our tools observe and take snapshots of it, in digital form, such as videos or sounds, however much of it is lost. Similarly for actions and information passing between humans, languages are used as a written form of communication. Traditionally, Machine Learning models have been unimodal (i.e. rgb -> semantic or text -> sentiment_class). Recent trends go towards bi-modality, where images and text are learned together, however, in order to truly understand the world, we need to integrate all these independent modalities. In this work we try to combine as many visual modalities as we can using little to no human supervision. In order to do this, we use pre-trained experts and procedural combinations between them on top of raw videos using a fully autonomous data-pipeline, which we also open-source. We then make use of PHG-MAE, a model specifically designed to leverage multi-modal data. We show that this model which was efficiently distilled into a low-parameter (<1M) can have competitive results compared to models of ~300M parameters. We deploy this model and analyze the use-case of real-time semantic segmentation from handheld devices or webcams on commodity hardware. Finally, we deploy other off-the-shelf models using the same framework, such as DPT for near real-time depth estimation.

Can models generalize attribute knowledge across semantically and perceptually dissimilar categories? While prior work has addressed attribute prediction within narrow taxonomic or visually similar domains, it remains unclear whether current models can abstract attributes and apply them to conceptually distant categories. This work presents the first explicit evaluation for the robustness of the attribute prediction task under such conditions, testing whether models can correctly infer shared attributes between unrelated object types: e.g., identifying that the attribute “has four legs” is common to both “dogs” and “chairs”. To enable this evaluation, we introduce train-test split strategies that progressively reduce correlation between training and test sets, based on: LLM-driven semantic grouping, embedding similarity thresholding, embedding-based clustering, and supercategory-based partitioning using ground-truth labels. Results show a sharp drop in performance as the correlation between training and test categories decreases, indicating strong sensitivity to split design. Among the evaluated methods, clustering yields the most effective trade-off, reducing hidden correlations while preserving learnability. These findings offer new insights into the limitations of current representations and inform future benchmark construction for attribute reasoning.

Ensuring safety in human–robot interaction requires reliable detection in industrial environments, yet the prohibitive cost of data annotation remains a major obstacle. We present a novel and cost-effective framework for automatic data generation and labeling, enabling unsupervised learning with state-of-the-art super-resolution LiDAR systems. Our method leverages recent advances in mesh reconstruction and animated 3D assets to create realistic synthetic datasets, surpassing existing augmentation strategies in fidelity and scalability. Beyond this, we introduce a second self-supervised stage that iteratively refines the model and bridges the virtual-to-real domain gap. Evaluations on real-world scenarios show substantial gains in performance.

Sparse Autoencoders (SAEs) are widely employed for mechanistic interpretability and model steering. Within this context, steering is by design performed by means of decoding altered SAE intermediate representations. This procedure essentially rewrites the original activations as a weighted sum of decoder features. In contrast to existing literature, we forward an encoder-centric alternative to model steering which demonstrates a stronger cross-modal performance. We introduce S&P Top-K, a retraining-free and computationally lightweight Selection and Projection framework that identifies Top-K encoder features aligned with a sensitive attribute or behavior, optionally aggregates them into a single control axis, and computes an orthogonal projection to be subsequently applied directly in the model’s native embedding space. In vision-language models, it improves fairness metrics on CelebA and FairFace by up to 3.2 times over conventional SAE usage, and in large language models, it substantially reduces aggressiveness and sycophancy in Llama-3 8B Instruct, achieving up to 3.6 times gains over masked reconstruction. These findings suggest that encoder-centric interventions provide a general, efficient, and more effective mechanism for shaping model behavior at inference time than the traditional decoder-centric use of SAEs.

Novelty detection seeks to identify samples deviating from a known distribution, yet data shifts in a multitude of ways, and only a few consist of relevant changes. Aligned with out-of-distribution generalization literature, we advocate for a formal distinction between task-relevant semantic or content changes and irrelevant style changes. This distinction forms the basis for robust novelty detection, emphasizing the identification of semantic changes resilient to style distributional shifts. To this end, we introduce Stylist, a method that utilizes pretrained large-scale model representations to selectively discard environment-biased features. By computing per-feature scores based on feature distribution distances between environments, Stylist effectively eliminates features responsible for spurious correlations, enhancing novelty detection performance. Evaluations on adapted domain generalization datasets and a synthetic dataset demonstrate Stylist’s efficacy in improving novelty detection across diverse datasets with stylistic and content shifts. The code is available at https://github.com/bit-ml/Stylist.

This paper presents a vision-only autonomous flight system for small UAVs operating in controlled indoor environments. The system combines semantic segmentation with monocular depth estimation to enable obstacle avoidance, scene exploration, and autonomous safe landing operations without requiring GPS or expensive sensors such as LiDAR. A key innovation is an adaptive scale factor algorithm that converts non-metric monocular depth predictions into accurate metric distance measurements by leveraging semantic ground plane detection and camera intrinsic parameters, achieving a mean distance error of 14.4 cm. The approach uses a knowledge distillation framework where a color-based Support Vector Machine (SVM) teacher generates training data for a lightweight U-Net student network (1.6M parameters) capable of real-time semantic segmentation. For more complex environments, the SVM teacher can be replaced with a state-of-the-art segmentation model. Testing was conducted in a controlled 5x4 meter laboratory environment with eight cardboard obstacles simulating urban structures. Extensive validation across 30 flight tests in a real-world environment and 100 flight tests in a digital-twin environment demonstrates that the combined segmentation and depth approach increases the distance traveled during surveillance and reduces mission time while maintaining 100% success rates. The system is further optimized through end-to-end learning, where a compact student neural network learns complete flight policies from demonstration data generated by our best-performing method, achieving an 87.5% autonomous mission success rate. This work advances practical vision-based drone navigation in structured environments, demonstrating solutions for metric depth estimation and computational efficiency challenges that enable deployment on resource-constrained platforms.

Summary (the publisher’s abstract is not openly available): Journal extension of TeachText (ICCV 2021). It proposes a generalized cross-modal distillation approach that leverages multiple text encoders to provide an enhanced supervisory signal for text-video retrieval, extends the idea to the video side to reduce the number of modalities needed at test time without loss in performance, and reports state-of-the-art results on several video retrieval benchmarks with no additional computational overhead at test time.

Pulmonary embolism (PE) remains a significant cause of cardiovascular mortality, with untreated cases showing mortality rates of up to 30%. The evolution of computer-assisted detection (CAD) for PE has transformed dramatically over the past decades, progressing from simple pattern recognition to sophisticated deep learning approaches. Early CAD systems demonstrated modest performance, with sensitivity around 75% at 2–4 false positives per scan, whereas modern deep learning architectures achieve sensitivities of up to 92.9% at 0.15 false positives per scan. Significantly, the technological progression has evolved from basic patient-level classification to sophisticated voxel-level analysis. This review provides a comprehensive overview of the evolution of PE CAD systems, their clinical value, and future directions.

In the current era of Machine Learning, Transformers have become the de facto approach across a variety of domains, such as computer vision and natural language processing. Transformer-based solutions are the backbone of current state-of-the-art methods for language generation, image and video classification, segmentation, action and object recognition, among many others. Interestingly enough, while these state-of-the-art methods produce impressive results in their respective domains, the problem of understanding the relationship between vision and language is still beyond our reach. In this work, we propose a common ground between vision and language based on events in space and time in an explainable and programmatic way, to connect learning-based vision and language state of the art models and provide a solution to the long standing problem of describing videos in natural language. We validate that our algorithmic approach is able to generate coherent, rich and relevant textual descriptions on videos collected from a variety of datasets, using both standard metrics (e.g. Bleu, ROUGE) and the modern LLM-as-a-Jury approach.

Humans effortlessly identify objects by leveraging a rich understanding of the surrounding scene, including spatial relationships, material properties, and the co-occurrence of other objects. In contrast, most computational object recognition systems operate on isolated image regions, devoid of meaning in isolation, thus ignoring this vital contextual information. This paper argues for the critical role of context and introduces a novel framework for contextual object classification. We first construct a Geo-Semantic Contextual Graph (GSCG) from a single monocular image. This rich, structured representation is built by integrating a metric depth estimator with a unified panoptic and material segmentation model. The GSCG encodes objects as nodes with detailed geometric, chromatic, and material attributes, and their spatial relationships as edges. This explicit graph structure makes the model’s reasoning process inherently interpretable. We then propose a specialized graph-based classifier that aggregates features from a target object, its immediate neighbors, and the global scene context to predict its class. Through extensive ablation studies, we demonstrate that our context-aware model achieves a classification accuracy of 73.4%, dramatically outperforming context-agnostic versions (as low as 38.4%). Furthermore, our GSCG-based approach significantly surpasses strong baselines, including fine-tuned ResNet models (max 53.5%) and a state-of-the-art multimodal Large Language Model (LLM), Llama 4 Scout, which, even when given the full image alongside a detailed description of objects, maxes out at 42.3%. These results on COCO 2017 train/val splits highlight the superiority of explicitly structured and interpretable context for object recognition tasks.

Generating novel views from recorded videos is crucial for enabling autonomous UAV navigation. Recent advancements in neural rendering have facilitated the rapid development of methods capable of rendering new trajectories. However, these methods often fail to generalize well to regions far from the training data without an optimized flight path, leading to suboptimal reconstructions. We propose a self-supervised cyclic neural-analytic pipeline that combines high-quality neural rendering outputs with precise geometric insights from analytical methods. Our solution improves RGB and mesh reconstructions for novel view synthesis, especially in undersampled areas and regions that are completely different from the training dataset. We use an effective transformer-based architecture for image reconstruction to refine and adapt the synthesis process, enabling effective handling of novel, unseen poses without relying on extensive labeled datasets. Our findings demonstrate substantial improvements in rendering views of novel and also 3D reconstruction, which to the best of our knowledge is a first, setting a new standard for autonomous navigation in complex outdoor environments.

Pulmonary Embolisms (PE) represent a leading cause of cardiovascular death. While medical imaging, through computed tomographic pulmonary angiography (CTPA), represents the gold standard for PE diagnosis, it is still susceptible to misdiagnosis or significant diagnosis delays, which may be fatal for critical cases. Despite the recently demonstrated power of deep learning to bring a significant boost in performance in a wide range of medical imaging tasks, there are still very few published researches on automatic pulmonary embolism detection. Herein we introduce a deep learning based approach, which efficiently combines computer vision and deep neural networks for pulmonary embolism detection in CTPA. Our method features novel improvements along three orthogonal axes: 1) automatic detection of anatomical structures; 2) anatomical aware pretraining, and 3) a dual-hop deep neural net for PE detection. We obtain state-of-the-art results on the publicly available multicenter large-scale RSNA dataset.

The Rashômon Effect, applied in Explainable Machine Learning, refers to the disagreement between the explanations provided by various attribution explainers and to the dissimilarity across multiple explanations generated by a particular explainer for a single instance from the dataset (differences between feature importances and their associated signs and ranks), an undesirable outcome especially in sensitive domains such as healthcare or finance. We propose a method inspired from textual-case based reasoning for aligning explanations from various explainers in order to resolve the disagreement and dissimilarity problems. We iteratively generated a number of 100 explanations for each instance from six popular datasets, using three prevalent feature attribution explainers: LIME, Anchors and SHAP (with the variations Tree SHAP and Kernel SHAP) and consequently applied a global cluster-based aggregation strategy that quantifies alignment and reveals similarities and associations between explanations. We evaluated our method by weighting the $$:k$$ -NN algorithm with agreed feature overlap explanation weights and compared it to a non-weighted $$:k$$ -NN predictor, having as task binary classification. Also, we compared the results of the weighted $$:k$$ -NN algorithm using aggregated feature overlap explanation weights to the weighted $$:k$$ -NN algorithm using weights produced by a single explanation method (either LIME, SHAP or Anchors). Our global alignment method benefited the most from a hybridization with feature importance scores (information gain), that was essential for acquiring a more accurate estimate of disagreement, for enabling explainers to reach a consensus across multiple explanations and for supporting effective model learning through improved classification performance.

Solving fish segmentation in underwater videos, a real-world problem of great practical value in marine and aquaculture industry, is a challenging task due to the difficulty of the filming environment, poor visibility and limited existing annotated underwater fish data. In order to overcome these obstacles, we introduce a novel two stage unsupervised segmentation approach that requires no human annotations and combines artificially created and real images. Our method generates challenging synthetic training data, by placing virtual fish in real-world underwater habitats, after performing fish transformations such as Thin Plate Spline shape warping and color Histogram Matching, which realistically integrate synthetic fish into the backgrounds, making the generated images increasingly closer to the real world data with every stage of our approach. While we validate our unsupervised method on the popular DeepFish dataset, obtaining a performance close to a fully-supervised SoTA model, we further show its effectiveness on the specific case of salmon segmentation in underwater videos, for which we introduce DeepSalmon, the largest dataset of its kind in the literature (30 GB). Moreover, on both datasets we prove the capability of our approach to boost the performance of the fully-supervised SoTA model.

Modeling face-to-face communication in computer vision, which focuses on recognizing and analyzing nonverbal cues and behaviors during interactions, serves as the foundation for our proposed alternative to text-based Human-AI interaction. By leveraging nonverbal visual communication, through facial expressions, head and body movements, we aim to enhance engagement and capture the user’s attention through a novel improvisational element, that goes beyond mirroring gestures. Our goal is to track and analyze facial expressions, and other nonverbal cues in real-time, and use this information to build models that can predict and understand human behavior. Operating in real-time and requiring minimal computational resources, our approach signifies a major leap forward in making AI interactions more natural and accessible. We offer three different complementary approaches, based on retrieval, statistical, and deep learning techniques. A key novelty of our work is the integration of an artistic component atop an efficient human-computer interaction system, using art as a medium to transmit emotions. Our approach is not art-specific and can be adapted to various paintings, animations, and avatars. In our experiments, we compare state-of-the-art diffusion models as mediums for emotion translation in 2D, and our 3D avatar, Maia, that we introduce in this work, with not just facial movements but also body motions for a more natural and engaging experience. We demonstrate the effectiveness of our approach in translating AI-generated emotions into human-relatable expressions, through both human and automatic evaluation procedures, highlighting its potential to significantly enhance the naturalness and engagement of Human-AI interactions across various applications.

There is an increasing number of real-world problems in computer vision and machine learning requiring to take into consideration multiple interpretation layers (modalities or views) of the world and learn how they relate to each other. For example, in the case of Earth Observations from satellite data, it is important to be able to predict one observation layer (e.g. vegetation index) from other layers (e.g. water vapor, snow cover, temperature etc), in order to best understand how the Earth System functions and also be able to reliably predict information for one layer when the data is missing (e.g. due to measurement failure or error).

In recent years, Large Language Models (LLMs) have achieved almost human-like performance on various tasks. While some LLMs have been trained on multilingual data, most of the training data is in English. Hence, their performance in English greatly exceeds their performance in other languages. This document presents our approach to training and evaluating the first foundational and chat LLM specialized for Romanian.

Quantifying the gap between synthetic and real-world imagery is essential for improving both transformer-based models - that rely on large volumes of data - and datasets, especially in underexplored domains like aerial scene understanding where the potential impact is significant. This paper introduces a novel methodology for scene complexity assessment using Multi-Model Consensus Metric (MMCM) and depth-based structural metrics, enabling a robust evaluation of perceptual and structural disparities between domains. Our experimental analysis, utilizing real-world (Dronescapes) and synthetic (Skyscenes) datasets, demonstrates that real-world scenes generally exhibit higher consensus among state-of-the-art vision transformers, while synthetic scenes show greater variability and challenge model adaptability. The results underline the inherent complexities and domain gaps, emphasizing the need for enhanced simulation fidelity and model generalization. This work provides critical insights into the interplay between domain characteristics and model performance, offering a pathway for improved domain adaptation strategies in aerial scene understanding.

In recent years, Large Language Models (LLMs) have achieved almost human-like performance on various tasks. While some LLMs have been trained on multilingual data, most of the training data is in English; hence, their performance in English greatly exceeds other languages. To our knowledge, we are the first to collect and translate a large collection of texts, instructions, and benchmarks and train, evaluate, and release open-source LLMs tailored for Romanian. We evaluate our methods on four different categories, including academic benchmarks, MT-Bench (manually translated), and a professionally built historical, cultural, and social benchmark adapted to Romanian. We argue for the usefulness and high performance of RoLLMs by obtaining state-of-the-art results across the board. We publicly release all resources (i.e., data, training and evaluation code, models) to support and encourage research on Romanian LLMs while concurrently creating a generalizable recipe, adequate for other low or less-resourced languages.

We tackle the problem of robust novelty detection, where we aim to detect novelties in terms of semantic content while being invariant to changes in other, irrelevant factors. Specifically, we operate in a setup with multiple environments, where we determine the set of features that are associated more with the environments, rather than to the content relevant for the task. Thus, we propose a method that starts with a pretrained embedding and a multi-env setup and manages to rank the features based on their environment-focus. First, we compute a per-feature score based on the feature distribution variance between envs. Next, we show that by dropping the highly scored ones, we manage to remove spurious correlations and improve the overall performance by up to 6%, both in covariance and sub-population shift cases, both for a real and a synthetic benchmark, that we introduce for this task.

Artificial Intelligence makes great advances today and starts to bridge the gap between vision and language. However, we are still far from understanding, explaining and controlling explicitly the visual content from a linguistic perspective, because we still lack a common explainable representation between the two domains. In this work we come to address this limitation and propose the Graph of Events in Space and Time (GEST), by which we can represent, create and explain, both visual and linguistic stories. We provide a theoretical justification of our model and an experimental validation, which proves that GEST can bring a solid complementary value along powerful deep learning models. In particular, GEST can help improve at the content-level the generation of videos from text, by being easily incorporated into our novel video generation engine. Additionally, by using efficient graph matching techniques, the GEST graphs can also improve the comparisons between texts at the semantic level.

One of the essential human skills is the ability to seamlessly build an inner representation of the world. By exploiting this representation, humans are capable of easily finding consensus between visual, auditory and linguistic perspectives. In this work, we set out to understand and emulate this ability through an explicit representation for both vision and language - Graphs of Events in Space and Time (GEST). GEST alows us to measure the similarity between texts and videos in a semantic and fully explainable way, through graph matching. It also allows us to generate text and videos from a common representation that provides a well understood content. In this work we show that the graph matching similarity metrics based on GEST outperform classical text generation metrics and can also boost the performance of state of art, heavily trained metrics.

We propose JEDI, a multi-dataset semi-supervised learning method, which efficiently combines knowledge from multiple experts, learned on different datasets, to train and improve the performance of individual, per dataset, student models. Our approach achieves this by addressing two important problems in current machine learning research: generalization across datasets and limitations of supervised training due to scarcity of labeled data. We start with an arbitrary number of experts, pretrained on their own specific dataset, which form the initial set of student models. The teachers are immediately derived by concatenating the feature representations from the penultimate layers of the students. We then train all models in a student-teacher semi-supervised learning scenario until convergence. In our efficient approach, student-teacher training is carried out jointly and end-to-end, showing that both students and teachers improve their generalization capacity during training. We validate our approach on four video action recognition datasets. By simultaneously considering all datasets within a unified semi-supervised setting, we demonstrate significant improvements over the initial experts.

There are many ways of interpreting the world and they are highly interdependent. We exploit such complex dependencies and introduce a powerful multi-task hypergraph, in which every node is a task and different paths through the hypergraph reaching a given task become unsupervised teachers, by forming ensembles that learn to generate reliable pseudolabels for that task. Each hyperedge is part of an ensemble teacher for a given task and it is also a student of the self-supervised hypergraph system. We apply our model to one of the most important problems of our times, that of Earth Observation, which is highly multi-task and it often suffers from missing ground-truth data. By performing extensive experiments on the NASA NEO Dataset, spanning a period of 22 years, we demonstrate the value of our multi-task semi-supervised approach, by consistent improvements over strong baselines and recent work. We also show that the hypergraph can adapt unsupervised to gradual data distribution shifts and reliably recover, through its multi-task self-supervision process, the missing data for several observational layers for up to seven years.

We present a method for learning multiple scene representations given a small labeled set, by exploiting the relationships between such representations in the form of a multi-task hypergraph. We also show how we can use the hypergraph to improve a powerful pretrained VisTransformer model without any additional labeled data. In our hypergraph, each node is an interpretation layer (e.g., depth or segmentation) of the scene. Within each hyperedge, one or several input nodes predict the layer at the output node. Thus, each node could be an input node in some hyperedges and an output node in others. In this way, multiple paths can reach the same node, to form ensembles from which we obtain robust pseudolabels, which allow self-supervised learning in the hypergraph. We test different ensemble models and different types of hyperedges and show superior performance to other multi-task graph models in the field. We also introduce Dronescapes, a large video dataset captured with UAVs in different complex real-world scenes, with multiple representations, suitable for multi-task learning.

The task of generating novel views of real scenes is increasingly important nowadays when AI models become able to create realistic new worlds. In many practical applications, it is important for novel view synthesis methods to stay grounded in the physical world as much as possible, while also being able to imagine it from previously unseen views. While most current methods are developed and tested in virtual environments with small scenes and no errors in pose and depth information, we push the boundaries to the real-world domain of large scales in the new context of UAVs. Our algorithmic contributions are two folds. First, we manage to stay anchored in the real 3D world, by introducing an efficient multi-scale voxel carving method, which is able to accommodate significant noises in pose, depth, and illumination variations, while being able to reconstruct the view of the world from drastically different poses at test time. Second, our final high-resolution output is efficiently self-trained on data automatically generated by the voxel carving module, which gives it the flexibility to adapt efficiently to any scene. We demonstrated the effectiveness of our method on highly complex and large-scale scenes in real environments while outperforming the current state-of-the-art. Our code is publicly available: https://github.com/onorabil/MSVC.

In this paper we propose a three stages analysis of the evolution of Covid19 in Romania. There are two main issues when it comes to pandemic prediction. The first one is the fact that the numbers reported of infected and recovered are unreliable, however the number of deaths is more accurate. The second issue is that there were many factors which affected the evolution of the pandemic. In this paper we propose an analysis in three stages. The first stage is based on the classical SIR model which we do using a neural network. This provides a first set of daily parameters. In the second stage we propose a refinement of the SIR model in which we separate the deceased into a distinct category. By using the first estimate and a grid search, we give a daily estimation of the parameters. The third stage is used to define a notion of turning points (local extremes) for the parameters. We call a regime the time between these points. We outline a general way based on time varying parameters of SIRD to make predictions.

Analyzing the distribution shift of data is a growing research direction in nowadays Machine Learning (ML), leading to emerging new benchmarks that focus on providing a suitable scenario for studying the generalization properties of ML models. The existing benchmarks are focused on supervised learning, and to the best of our knowledge, there is none for unsupervised learning. Therefore, we introduce an unsupervised anomaly detection benchmark with data that shifts over time, built over Kyoto-2006+, a traffic dataset for network intrusion detection. This type of data meets the premise of shifting the input distribution: it covers a large time span ($10$ years), with naturally occurring changes over time (eg users modifying their behavior patterns, and software updates). We first highlight the non-stationary nature of the data, using a basic per-feature analysis, t-SNE, and an Optimal Transport approach for measuring the overall distribution distances between years. Next, we propose AnoShift, a protocol splitting the data in IID, NEAR, and FAR testing splits. We validate the performance degradation over time with diverse models, ranging from classical approaches to deep learning. Finally, we show that by acknowledging the distribution shift problem and properly addressing it, the performance can be improved compared to the classical training which assumes independent and identically distributed data (on average, by up to $3%$ for our approach). Dataset and code are available at https://github.com/bit-ml/AnoShift/.

We introduce a formalization and benchmark for the unsupervised anomaly detection task in the distribution-shift scenario. Our work builds upon the iWildCam dataset, and, to the best of our knowledge, we are the first to propose such an approach for visual data. We empirically validate that environment-aware methods perform better in such cases when compared with the basic Empirical Risk Minimization (ERM). We next propose an extension for generating positive samples for contrastive methods that considers the environment labels when training, improving the ERM baseline score by 8.7%.

We propose a dual system for unsupervised object segmentation in video, which brings together two modules with complementary properties: a space-time graph that discovers objects in videos and a deep network that learns powerful object features. The system uses an iterative knowledge exchange policy. A novel spectral space-time clustering process on the graph produces unsupervised segmentation masks passed to the network as pseudo-labels. The net learns to segment in single frames what the graph discovers in video and passes back to the graph strong image-level features that improve its node-level features in the next iteration. Knowledge is exchanged for several cycles until convergence. The graph has one node per each video pixel, but the object discovery is fast. It uses a novel power iteration algorithm computing the main space-time cluster as the principal eigenvector of a special Feature-Motion matrix without actually computing the matrix. The thorough experimental analysis validates our theoretical claims and proves the effectiveness of the cyclical knowledge exchange. We also perform experiments on the supervised scenario, incorporating features pretrained with human supervision. We achieve state-of-the-art level on unsupervised and supervised scenarios on four challenging datasets: DAVIS, SegTrack, YouTube-Objects, and DAVSOD.

One of the main drivers of the recent advances in authorship verification is the PAN large-scale authorship dataset. Despite generating significant progress in the field, inconsistent performance differences between the closed and open test sets have been reported. To this end, we improve the experimental setup by proposing five new public splits over the PAN dataset, specifically designed to isolate and identify biases related to the text topic and to the author’s writing style. We evaluate several BERT-like baselines on these splits, showing that such models are competitive with authorship verification state-of-the-art methods. Furthermore, using explainable AI, we find that these baselines are biased towards named entities. We show that models trained without the named entities obtain better results and generalize better when tested on DarkReddit, our new dataset for authorship verification.

We propose an efficient method for unsupervised learning of metric depth estimation from a single image in the context of unconstrained videos captured from UAVs. We combine the accuracy of an analytical solution based on odometry with the power of deep learning. First, we show how to correct the noisy odometric measurements by optimizing the alignment between the derotated optical flow and the projected linear speed in the image. Then, we detail an analytical depth estimation method based on optical flow and corrected camera velocities. Subsequently, the improved depth and camera veloc-ities obtained analytically are used, as additional cost terms, for training our novel unsupervised learning architecture for metric depth estimation. We extensively test on a recent UAV dataset, which we significantly extend by adding completely novel scenes. We outperform by significant margins different kinds of state-of-the-art approaches, ranging from analytical and unsupervised solutions to transformer-based architectures that require heavy computation and pre-training. The resulting algorithm could be deployed on embedded devices, being a good candidate for practical robotics use cases, such as obstacle avoidance and safe landing for UAV s.

Abnormal event detection in video is a complex computer vision problem that has attracted significant attention in recent years. The complexity of the task arises from the commonly-adopted definition of an abnormal event, that is, a rarely occurring event that typically depends on the surrounding context. Following the standard formulation of abnormal event detection as outlier detection, we propose a background-agnostic framework that learns from training videos containing only normal events. Our framework is composed of an object detector, a set of appearance and motion auto-encoders, and a set of classifiers. Since our framework only looks at object detections, it can be applied to different scenes, provided that normal events are defined identically across scenes and that the single main factor of variation is the background. To overcome the lack of abnormal data during training, we propose an adversarial learning strategy for the auto-encoders. We create a scene-agnostic set of out-of-domain pseudo-abnormal examples, which are correctly reconstructed by the auto-encoders before applying gradient ascent on the pseudo-abnormal examples. We further utilize the pseudo-abnormal examples to serve as abnormal examples when training appearance-based and motion-based binary classifiers to discriminate between normal and abnormal latent features and reconstructions. We compare our framework with the state-of-the-art methods on four benchmark data sets, using various evaluation metrics. Compared to existing methods, the empirical results indicate that our approach achieves favorable performance on all data sets. In addition, we provide region-based and track-based annotations for two large-scale abnormal event detection data sets from the literature, namely ShanghaiTech and Subway.

Phobia is an anxiety disorder that affects 13% of the world’s population and it is manifested through an extreme and irrational fear toward objects or situations. A lot of research has been done on studying the contributing factors to the onset, development, and maintenance of phobias, underlying cognitive and behavioral processes, physical manifestation, and treatment methods. In this chapter, we present a new automated approach to phobias therapy, based on the integration of virtual reality technology, artificial intelligence, and affective computing—the ability of computational machines to recognize, adapt, and respond intelligently to human emotions. First, we introduce the results obtained by applying various machine learning algorithms for classifying the six basic emotions—anger, disgust, fear, joy, sadness, and surprise—based on the physiological recordings and self-reported ratings from the DEAP database. Then we present the stages of development and evaluation of a virtual environment for treating acrophobia that relies on gradual exposure to stimuli, accompanied by physiological signals monitoring. In a pilot experiment, we proposed a model for automatically adapting exposure scenarios according to the users’ emotional states, which has been tested on four acrophobic subjects. Here, the results obtained pass through a critical analysis in which we discuss the progress of the research and current limitations. Finally, we introduce the conceptual design of a system for phobias treatment that can be used as a therapy tool in clinical setups and also for home training, integrating virtual reality, machine learning, gamification techniques, biophysical data recording, emotion recognition, and automatic scenario adaptation. The system uses an intelligent virtual therapist that can understand emotions based on physiological signals, provides encouragement, and changes the level of exposure according to the user’s affective states.

Anomaly detection in video is a challenging computer vision problem. Due to the lack of anomalous events at training time, anomaly detection requires the design of learning methods without full supervision. In this paper, we approach anomalous event detection in video through self-supervised and multi-task learning at the object level. We first utilize a pre-trained detector to detect objects. Then, we train a 3D convolutional neural network to produce discriminative anomaly-specific information by jointly learning multiple proxy tasks: three self-supervised and one based on knowledge distillation. The self-supervised tasks are: (i) discrimination of forward/backward moving objects (arrow of time), (ii) discrimination of objects in consecutive/intermittent frames (motion irregularity) and (iii) reconstruction of object-specific appearance information. The knowledge distillation task takes into account both classification and detection information, generating large prediction discrepancies between teacher and student models when anomalies occur. To the best of our knowledge, we are the first to approach anomalous event detection in video as a multi-task learning problem, integrating multiple self-supervised and knowledge distillation proxy tasks in a single architecture. Our lightweight architecture outperforms the state-of-the-art methods on three benchmarks: Avenue, ShanghaiTech and UCSD Ped2. Additionally, we perform an ablation study demonstrating the importance of integrating self-supervised learning and normality-specific distillation in a multi-task learning setting.

Leveraging deep learning models for Anomaly Detection (AD) has seen widespread use in recent years due to superior performances over traditional methods. Recent deep methods for anomalies in images learn better features of normality in an end-to-end self-supervised setting. These methods train a model to discriminate between different transformations applied to visual data and then use the output to compute an anomaly score. We use this approach for AD in text, by introducing a novel pretext task on text sequences. We learn our DATE model end-to-end, enforcing two independent and complementary self-supervision signals, one at the token-level and one at the sequence-level. Under this new task formulation, we show strong quantitative and qualitative results on the 20Newsgroups and AG News datasets. In the semi-supervised setting, we outperform state-of-the-art results by +13.5% and +6.9%, respectively (AUROC). In the unsupervised configuration, DATE surpasses all other methods even when 10% of its training data is contaminated with outliers (compared with 0% for the others).

Estimating precise metric depth is an essential task for UAV navigation, which is very difficult to learn unsupervised without access to odometry. At the same time, depth recovery from kinematics and optical flow is mathematically precise, but less numerically stable and robust, especially in the focus of expansion areas. We propose a model that combines the analytical, vision-with-odometry approach, with deep unsupervised learning, into a single formulation for metric depth estimation, which is both fast and accurate. The two pathways – analytical and data-driven – form a robust ensemble, which provides supervisory signal to a single deep net that distills the consensus between scene geometry, pose, kinematics, camera intrinsics and the input RGB. The distilled net has low runtime and memory costs, being suitable for embedded devices. We validate our results against an off-the-shelf SfM-based solution. We also introduce a new real-world dataset of almost 20 minutes of continuous UAV flight, on which we demonstrate better accuracy and capabilities than the deep learning and analytical approaches.

Graph Neural Networks are perfectly suited to capture latent interactions between various entities in the spatio-temporal domain (e.g. videos). However, when an explicit structure is not available, it is not obvious what atomic elements should be represented as nodes. Current works generally use pre-trained object detectors or fixed, predefined regions to extract graph nodes. Improving upon this, our proposed model learns nodes that dynamically attach to well-delimited salient regions, which are relevant for a higher-level task, without using any object-level supervision. Constructing these localized, adaptive nodes gives our model inductive bias towards object-centric representations and we show that it discovers regions that are well correlated with objects in the video. In extensive ablation studies and experiments on two challenging datasets, we show superior performance to previous graph neural networks models for video classification.

When driving, people make decisions based on current traffic as well as their desired route. They have a mental map of known routes and are often able to navigate without needing directions. Current published self-driving models improve their performances when using additional GPS information. Here we aim to push forward self-driving research and perform route planning even in the complete absence of GPS at inference time. Our system learns to predict in real-time vehicle’s current location and future trajectory, on a known map, given only the raw video stream and the final destination. Trajectories consist of instant steering commands that depend on present traffic, as well as longer-term navigation decisions towards a specific destination. Along with our novel proposed approach to localization and navigation from visual data, we also introduce a novel large dataset in an urban environment, which consists of video and GPS streams collected with a smartphone while driving. The GPS is automatically processed to obtain supervision labels and to create an analytical representation of the traversed map. In tests, our solution outperforms published state of the art methods on visual localization and steering and provides reliable navigation assistance between any two known locations. We also show that our system can adapt to short and long-term changes in weather conditions or the structure of the urban environment. We make the entire dataset and the code publicly available.

The human ability to synchronize the feedback from all their senses inspired recent works in multi-task and multi-modal learning. While these works rely on expensive supervision, our multi-task graph requires only pseudo-labels from expert models. Every graph node represents a task, and each edge learns between tasks transformations. Once initialized, the graph learns self-supervised, based on a novel consensus shift algorithm that intelligently exploits the agreement between graph pathways to generate new pseudo-labels for the next learning cycle. We demonstrate significant improvement from one unsupervised learning iteration to the next, outperforming related recent methods in extensive multi-task learning experiments on two challenging datasets. Our code is available at https://github.com/bit-ml/cshift.

We address the challenging problem of semi-supervised learning in the context of multiple visual interpretations of the world by finding consensus in a graph of neural networks. Each graph node is a scene interpretation layer, while each edge is a deep net that transforms one layer at one node into another from a different node. During the supervised phase edge networks are trained independently. During the next unsupervised stage edge nets are trained on the pseudo-ground truth provided by consensus among multiple paths that reach the nets’ start and end nodes. These paths act as ensemble teachers for any given edge and strong consensus is used for high-confidence supervisory signal. The unsupervised learning process is repeated over several generations, in which each edge becomes a “student” and also part of different ensemble “teachers” for training other students. By optimizing such consensus between different paths, the graph reaches consistency and robustness over multiple interpretations and generations, in the face of unknown labels. We give theoretical justifications of the proposed idea and validate it on a large dataset. We show how prediction of different representations such as depth, semantic segmentation, surface normals and pose from RGB input could be effectively learned through self-supervised consensus in our graph. We also compare to state-of-the-art methods for multi-task and semi-supervised learning and show superior performance.

In recent years, considerable progress on the task of text-video retrieval has been achieved by leveraging large-scale pretraining on visual and audio datasets to construct powerful video encoders. By contrast, despite the natural symmetry, the design of effective algorithms for exploiting large-scale language pretraining remains under-explored. In this work, we are the first to investigate the design of such algorithms and propose a novel generalized distillation method, TeachText, which leverages complementary cues from multiple text encoders to provide an enhanced supervisory signal to the retrieval model. Moreover, we extend our method to video side modalities and show that we can effectively reduce the number of used modalities at test time without compromising performance. Our approach advances the state of the art on several video retrieval benchmarks by a significant margin and adds no computational overhead at test time. Last but not least, we show an effective application of our method for eliminating noise from retrieval datasets. Code and data can be found at https://www.robots.ox.ac.uk/~vgg/research/teachtext/.

We formulate object segmentation in video as a graph partitioning problem in space and time, in which nodes are pixels and their relations form local neighborhoods. We claim that the strongest cluster in this pixel-level graph represents the salient object segmentation. We compute the main cluster using a novel and fast 3D filtering technique that finds the spectral clustering solution, namely the principal eigenvector of the graph’s adjacency matrix, without building the matrix explicitly - which would be intractable. Our method is based on the power iteration for finding the principal eigenvector of a matrix, which we prove is equivalent to performing a specific set of 3D convolutions in the space-time feature volume. This allows us to avoid creating the matrix and have a fast parallel implementation on GPU. We show that our method is much faster than classical power iteration applied directly on the adjacency matrix. Different from other works, ours is dedicated to preserving object consistency in space and time at the level of pixels. For that, it requires powerful pixel-wise features at the frame level. This makes it perfectly suitable for incorporating the output of a backbone network or other methods and fast-improving over their solution without supervision. In experiments, we obtain consistent improvement, with the same set of hyper-parameters, over the top state of the art methods on DAVIS-2016 dataset, both in unsupervised and semi-supervised tasks. We also achieve top results on the well-known SegTrackv2 dataset.

Automatically describing videos in natural language is an ambitious problem, which could bridge our understanding of vision and language.We propose a hierarchical approach, by first generating video descriptions as sequences of simple sentences, followed at the next level by a more complex and fluent description in natural language.While the simple sentences describe simple actions in the form of (subject, verb, object), the second-level paragraph descriptions, indirectly using information from the first-level description, presents the visual content in a more compact, coherent and semantically rich manner.To this end, we introduce the first video dataset in the literature that is annotated with captions at two levels of linguistic complexity.We perform extensive tests that demonstrate that our hierarchical linguistic representation, from simple to complex language, allows us to train a two-stage network that is able to generate significantly more complex paragraphs than current one-stage approaches.

In this paper we propose a three stages analysis of the evolution of Covid19 in Romania. There are two main issues when it comes to pandemic prediction. The first one is the fact that the numbers reported of infected and recovered are unreliable, however the number of deaths is more accurate. The second issue is that there were many factors which affected the evolution of the pandemic. In this paper we propose an analysis in three stages. The first stage is based on the classical SIR model which we do using a neural network. This provides a first set of daily parameters. In the second stage we propose a refinement of the SIR model in which we separate the deceased into a distinct category. By using the first estimate and a grid search, we give a daily estimation of the parameters. The third stage is used to define a notion of turning points (local extremes) for the parameters. We call a regime the time between these points. We outline a general way based on time varying parameters of SIRD to make predictions.

Analysing and understanding the transmission and evolution of the COVID-19 pandemic is mandatory to be able to design the best social and medical policies, foresee their outcomes and deal with all the subsequent socio-economic effects. We address this important problem from a computational and machine learning perspective. More specifically, we want to statistically estimate all the relevant parameters for the new coronavirus COVID-19, such as the reproduction number, fatality rate or length of infectiousness period, based on Romanian patients, as well as be able to predict future outcomes. This endeavor is important, since it is well known that these factors vary across the globe, and might be dependent on many causes, including social, medical, age and genetic factors. We use a recently published improved version of SEIR, which is the classic, established model for infectious diseases. We want to infer all the parameters of the model, which govern the evolution of the pandemic in Romania, based on the only reliable, true measurement, which is the number of deaths. Once the model parameters are estimated, we are able to predict all the other relevant measures, such as the number of exposed and infectious people. To this end, we propose a self-supervised approach to train a deep convolutional network to guess the correct set of Modified-SEIR model parameters, given the observed number of daily fatalities. Then, we refine the solution with a stochastic coordinate descent approach. We compare our deep learning optimization scheme with the classic grid search approach and show great improvement in both computational time and prediction accuracy. We find an optimistic result in the case fatality rate for Romania which may be around 0.3% and we also demonstrate that our model is able to correctly predict the number of daily fatalities for up to three weeks in the future.

In this paper, we investigate various machine learning classifiers used in our Virtual Reality (VR) system for treating acrophobia. The system automatically estimates fear level based on multimodal sensory data and a self-reported emotion assessment. There are two modalities of expressing fear ratings: the 2-choice scale, where 0 represents relaxation and 1 stands for fear; and the 4-choice scale, with the following correspondence: 0-relaxation, 1-low fear, 2-medium fear and 3-high fear. A set of features was extracted from the sensory signals using various metrics that quantify brain (electroencephalogram-EEG) and physiological linear and non-linear dynamics (Heart Rate-HR and Galvanic Skin Response-GSR). The novelty consists in the automatic adaptation of exposure scenario according to the subject’s affective state. We acquired data from acrophobic subjects who had undergone an in vivo pre-therapy exposure session, followed by a Virtual Reality therapy and an in vivo evaluation procedure. Various machine and deep learning classifiers were implemented and tested, with and without feature selection, in both a user-dependent and user-independent fashion. The results showed a very high cross-validation accuracy on the training set and good test accuracies, ranging from 42.5% to 89.5%. The most important features of fear level classification were GSR, HR and the values of the EEG in the beta frequency range. For determining the next exposure scenario, a dominant role was played by the target fear level, a parameter computed by taking into account the patient’s estimated fear level.

The task of detecting whether a person wears a face mask from speech is useful in modelling speech in forensic investigations, communication between surgeons or people protecting themselves against infectious diseases such as COVID-19. In this paper, we propose a novel data augmentation approach for mask detection from speech. Our approach is based on (i) training Generative Adversarial Networks (GANs) with cycle-consistency loss to translate unpaired utterances between two classes (with mask and without mask), and on (ii) generating new training utterances using the cycle-consistent GANs, assigning opposite labels to each translated utterance. Original and translated utterances are converted into spectrograms which are provided as input to a set of ResNet neural networks with various depths. The networks are combined into an ensemble through a Support Vector Machines (SVM) classifier. With this system, we participated in the Mask Sub-Challenge (MSC) of the INTERSPEECH 2020 Computational Paralinguistics Challenge, surpassing the baseline proposed by the organizers by 2.8%. Our data augmentation technique provided a performance boost of 0.9% on the private test set. Furthermore, we show that our data augmentation approach yields better results than other baseline and state-of-the-art augmentation methods.

Summary (the publisher’s abstract is not openly available): Within a holonic, agent-based architecture for virtual-reality phobia therapy, this work classifies users’ fear levels from physiological signals (EEG, galvanic skin response, heart rate) recorded during VR exposure, comparing several machine-learning and deep-learning classifiers in order to automatically adapt the exposure intensity to the patient’s current affective state.

In this paper, we propose a deep learning approach for smartphone user identification based on analyzing motion signals recorded by the accelerometer and the gyroscope, during a single tap gesture performed by the user on the screen. We transform the discrete 3-axis signals from the motion sensors into a gray-scale image representation which is provided as input to a convolutional neural network (CNN) that is pre-trained for multi-class user classification. In the pre-training stage, we benefit from different users and multiple samples per user. After pre-training, we use our CNN as feature extractor, generating an embedding associated to each single tap on the screen. The resulting embeddings are used to train a binary Support Vector Machines (SVM) model in a few-shot user identification setting, i.e. requiring only 20 taps on the screen during the registration phase. We compare our identification system based on CNN features with two baseline systems, one that employs handcrafted features and another that employs recurrent neural networks (RNN) features. All systems are based on the same classifier, namely SVM. To pre-train the CNN and the RNN models for multi-class user classification, we use a different set of users than the set used for few-shot user identification, ensuring a realistic scenario. The empirical results demonstrate that our CNN model yields a top accuracy of 90.75% in multi-class user classification and a top accuracy of 96.72% in few-shot user identification. We thus believe that our system is ready for practical use, having a better generalization capacity than both baselines. We also conduct experiments showing that the binary SVM provides better results than the one-class SVM, although the negative samples added during training do not belong to the attackers (known only at test time).

Computed Tomography (CT) scanners that are commonly-used in hospitals and medical centers nowadays produce low-resolution images, e.g. one voxel in the image corresponds to at most one-cubic millimeter of tissue. In order to accurately segment tumors and make treatment plans, radiologists and oncologists need CT scans of higher resolution. The same problem appears in Magnetic Resonance Imaging (MRI). In this paper, we propose an approach for the single-image super-resolution of 3D CT or MRI scans. Our method is based on deep convolutional neural networks (CNNs) composed of 10 convolutional layers and an intermediate upscaling layer that is placed after the first 6 convolutional layers. Our first CNN, which increases the resolution on two axes (width and height), is followed by a second CNN, which increases the resolution on the third axis (depth). Different from other methods, we compute the loss with respect to the ground-truth high-resolution image right after the upscaling layer, in addition to computing the loss after the last convolutional layer. The intermediate loss forces our network to produce a better output, closer to the ground-truth. A widely-used approach to obtain sharp results is to add Gaussian blur using a fixed standard deviation. In order to avoid overfitting to a fixed standard deviation, we apply Gaussian smoothing with various standard deviations, unlike other approaches. We evaluate the proposed method in the context of 2D and 3D super-resolution of CT and MRI scans from two databases, comparing it to related works from the literature and baselines based on various interpolation schemes, using 2× and 4× scaling factors. The empirical study shows that our approach attains superior results to all other methods. Moreover, our subjective image quality assessment by human observers reveals that both doctors and regular annotators chose our method in favor of Lanczos interpolation in 97.55% cases for an upscaling factor of 2× and in 96.69% cases for an upscaling factor of 4×. In order to allow others to reproduce our state-of-the-art results, we provide our code as open source at https://github.com/lilygeorgescu/3d-super-res-cnn.

Graph Neural Networks are perfectly suited to capture latent interactions occurring in the spatio-temporal domain. But when an explicit structure is not available, as in the visual domain, it is not obvious what atomic elements should be represented as nodes. They should depend on the context and the kinds of relations that we are interested in. We are focusing on modeling relations between instances by proposing a method that takes advantage of the locality assumption to create nodes that are clearly localised in space. Current works are using external object detectors or fixed regions to extract features corresponding to graph nodes, while we propose a module for generating the regions associated with each node dynamically, without explicit object-level supervision. Conditioned on the input, for each node we predict the location and size of a region and use them to pool node features using a differentiable mechanism. Constructing these localised, adaptive nodes makes our model biased towards object-centric representations and we show that it improves the modeling of visual interactions. By relying on a few localized nodes, our method learns to focus on salient regions leading to a more explainable model. Our model achieves superior results on video classification tasks involving instance interactions.

Emotions constitute an indispensable component of our everyday life. They consist of conscious mental reactions towards objects or situations and are associated with various physiological, behavioral, and cognitive changes. In this paper, we propose a comparative analysis between different machine learning and deep learning techniques, with and without feature selection, for binarily classifying the six basic emotions, namely anger, disgust, fear, joy, sadness, and surprise, into two symmetrical categorical classes (emotion and no emotion), using the physiological recordings and subjective ratings of valence, arousal, and dominance from the DEAP (Dataset for Emotion Analysis using EEG, Physiological and Video Signals) database. The results showed that the maximum classification accuracies for each emotion were: anger: 98.02%, joy:100%, surprise: 96%, disgust: 95%, fear: 90.75%, and sadness: 90.08%. In the case of four emotions (anger, disgust, fear, and sadness), the classification accuracies were higher without feature selection. Our approach to emotion classification has future applicability in the field of affective computing, which includes all the methods used for the automatic assessment of emotions and their applications in healthcare, education, marketing, website personalization, recommender systems, video games, and social media.

Summary (the publisher’s abstract is not openly available): eTher is an assistive virtual agent for acrophobia therapy in a VR mountain landscape with a cable-car ride. The agent talks to the user, adapts its voice pitch, tempo and volume to the patient’s emotional state estimated from biophysical signals relative to a resting baseline, and automatically lowers the exposure level when relaxation techniques do not work. A pilot study with 10 users showed faster task completion and significantly improved relaxation when assisted by the agent.

Despite the significant advances in recent years, Generative Adversarial Networks (GANs) are still notoriously hard to train. In this paper, we propose three novel curriculum learning strategies for training GANs. All strategies are first based on ranking the training images by their difficulty scores, which are estimated by a state-of-the-art image difficulty predictor. Our first strategy is to divide images into gradually more difficult batches. Our second strategy introduces a novel curriculum loss function for the discriminator that takes into account the difficulty scores of the real images. Our third strategy is based on sampling from an evolving distribution, which favors the easier images during the initial training stages and gradually converges to a uniform distribution, in which samples are equally likely, regardless of difficulty. We compare our curriculum learning strategies with the classic training procedure on two tasks: image generation and image translation. Our experiments indicate that all strategies provide faster convergence and superior results. For example, our best curriculum learning strategy applied on spectrally normalized GANs (SNGANs) fooled human annotators in thinking that generated CIFAR-like images are real in 25.0% of the presented cases, while the SNGANs trained using the classic procedure fooled the annotators in only 18.4% cases. Similarly, in image translation, the human annotators preferred the images produced by the Cycle-consistent GAN (CycleGAN) trained using curriculum learning in 40.5% cases and those produced by CycleGAN based on classic training in only 19.8% cases, 39.7% cases being labeled as ties.

Automatically detecting vital signs in videos, such as the estimation of heart and respiration rates, is a challenging research problem in computer vision with important applications in the medical field. One of the key difficulties in tackling this task is the lack of sufficient supervised training data, which severely limits the use of powerful deep neural networks. In this paper we address this limitation through a novel deep learning approach, in which a recurrent deep neural network is trained to detect vital signs in the infrared thermal domain from purely synthetic data. What is most surprising is that our novel method for synthetic training data generation is general, relatively simple and uses almost no prior medical domain knowledge. Moreover, our system, which is trained in a purely automatic manner and needs no human annotation, also learns to predict the respiration or heart intensity signal for each moment in time and to detect the region of interest that is most relevant for the given task, e.g. the nose area in the case of respiration. We test the effectiveness of our proposed system on the recent LCAS dataset and obtain state-of-the-art results.

This paper proposes a protocol for the acquisition and processing of biophysical signals in virtual reality applications, particularly in phobia therapy experiments. This protocol aims to ensure that the measurement and processing phases are performed effectively, to obtain clean data that can be used to estimate the users’ anxiety levels. The protocol has been designed after analyzing the experimental data of seven subjects who have been exposed to heights in a virtual reality environment. The subjects’ level of anxiety has been estimated based on the real-time evaluation of a nonlinear function that has as parameters various features extracted from the biophysical signals. The highest classification accuracy was obtained using a combination of seven heart rate and electrodermal activity features in the time domain and frequency domain.

We propose a method at the intersection of Computer Vision and Computer Graphics fields, which automatically generates RGBD images using neural networks, based on previously seen and synchronized video, depth and pose signals. Since the models must be able to reconstruct both texture (RGB) and structure (Depth), it creates an implicit representation of the scene, as opposed to explicit ones, such as meshes or point clouds. The process can be thought of as neural rendering, where we obtain a function f : Pose -> RGBD, which we can use to navigate through the generated scene, similarly to graphics simulations. We introduce two new datasets, one based on synthetic data with full ground truth information, while the other one being recorded from a drone flight in an university campus, using only video and GPS signals. Finally, we propose a fully unsupervised method of generating datasets from videos alone, in order to train the Pose2RGBD networks. Code and datasets are available at:: https://gitlab.com/mihaicristianpirvu/pose2rgbd.

Classifying visual information is an apparently simple and effortless task in our everyday routine, but can we automatically predict what we see from signals emitted by the brain? While other researchers have already attempted to answer this question, we are the first to show that a commercially available BCI could be effectively used for visual image classification in real-world scenarios – when testing takes place at a completely different time than training data collection. The task is difficult, as it requires relating the noisy and low-level EEG signals to complex and highly semantic visual categories. In this paper, we propose different learning approaches and show that simpler classifiers such as Ridge Regression with Gabor filtering of the input EEG signal could be more effective than the powerful Long Short Term Memory Networks and Convolutional Neural Networks in this case of limited and noisy training data. We analyzed the importance of each electrode for the visual classification task and noticed that the sensors with the highest accuracy were the ones that recorded brain activity from regions known to be correlated more with higher level recognition and cognitive processes and less to lower-level visual signal processing. The result is also in accordance with research in computer vision with deep neural networks, which shows that semantic visual features are learned only at higher levels of neural depth. While EEG signals are weaker by themselves for the task of visual classification, we demonstrate that they could be powerful when combined with deep visual features extracted from the image, improving performance from 91% to over 97% in a multi-class recognition setting. Our tests show that EEG input brings additional information that is not learned by artificial deep networks on the given image training set. Thus, a commercially available BCI could be effectively used in conjunction with a deep learning based vision system to form together a stronger visual recognition system that is suitable for real-world applications.

Summary (the publisher’s abstract is not openly available): Presents a real-time pipeline for detecting multiple targets in the wild from a moving 360° panoramic camera, addressing the distortions and wide field of view of omnidirectional imagery for embedded, on-board use (Embedded Vision Workshop, ECCV 2020).

Deep pre-trained language models tend to become ubiquitous in the field of Natural Language Processing (NLP).These models learn contextualized representations by using a huge amount of unlabeled text data and obtain state of the art results on a multitude of NLP tasks, by enabling efficient transfer learning.For other languages besides English, there are limited options of such models, most of which are trained only on multi-lingual corpora.In this paper we introduce a Romanian-only pre-trained BERT model -RoBERT -and compare it with different multilingual models on seven Romanian specific NLP tasks grouped into three categories, namely: sentiment analysis, dialect and cross-dialect topic identification, and diacritics restoration.Our model surpasses the multi-lingual models, as well as a another mono-lingual implementation of BERT, on all tasks.

Semantic segmentation is a crucial task for robot navigation and safety. However, current supervised methods require a large amount of pixelwise annotations to yield accurate results. Labeling is a tedious and time consuming process that has hampered progress in low altitude UAV applications. This paper makes an important step towards automatic annotation by introducing SegProp, a novel iterative flow-based method, with a direct connection to spectral clustering in space and time, to propagate the semantic labels to frames that lack human annotations. The labels are further used in semi-supervised learning scenarios. Motivated by the lack of a large video aerial dataset, we also introduce Ruralscapes, a new dataset with high resolution (4K) images and manually-annotated dense labels every 50 frames - the largest of its kind, to the best of our knowledge. Our novel SegProp automatically annotates the remaining unlabeled 98% of frames with an accuracy exceeding 90% (F-measure), significantly outperforming other state-of-the-art label propagation methods. Moreover, when integrating other methods as modules inside SegProp’s iterative label propagation loop, we achieve a significant boost over the baseline labels. Finally, we test SegProp in a full semi-supervised setting: we train several state-of-the-art deep neural networks on the SegProp-automatically-labeled training frames and test them on completely novel videos. We convincingly demonstrate, every time, a significant improvement over the supervised scenario.

We propose an object tracking method, SFTrack++, that smoothly learns to preserve the tracked object consistency over space and time dimensions by taking a spectral clustering approach over the graph of pixels from the video, using a fast 3D filtering formulation for finding the principal eigenvector of this graph’s adjacency matrix. To better capture complex aspects of the tracked object, we enrich our formulation to multi-channel inputs, which permit different points of view for the same input. The channel inputs are in our experiments, the output of multiple tracking methods. After combining them, instead of relying only on hidden layers representations to predict a good tracking bounding box, we explicitly learn an intermediate, more refined one, namely the segmentation map of the tracked object. This prevents the rough common bounding box approach to introduce noise and distractors in the learning process. We test our method, SFTrack++, on five tracking benchmarks: OTB, UAV, NFS, GOT-10k, and TrackingNet, using five top trackers as input. Our experimental results validate the pre-registered hypothesis. We obtain consistent and robust results, competitive on the three traditional benchmarks (OTB, UAV, NFS) and significantly on top of others (by over $1.1%$ on accuracy) on GOT-10k and TrackingNet, which are newer, larger, and more varied datasets.

Nowadays, commonly-used authentication systems for mobile device users, e.g. password checking, face recognition or fingerprint scanning, are susceptible to various kinds of attacks. In order to prevent some of the possible attacks, these explicit authentication systems can be enhanced by considering a two-factor authentication scheme, in which the second factor is an implicit authentication system based on analyzing motion sensor data captured by accelerometers or gyroscopes. In order to avoid any additional burdens to the user, the registration process of the implicit authentication system must be performed quickly, i.e. the number of data samples collected from the user is typically small. In the context of designing a machine learning model for implicit user authentication based on motion signals, data augmentation can play an important role. In this paper, we study several data augmentation techniques in the quest of finding useful augmentation methods for motion sensor data. We propose a set of four research questions related to data augmentation in the context of few-shot user identification based on motion sensor signals. We conduct experiments on a benchmark data set, using two deep learning architectures, convolutional neural networks and Long Short-Term Memory networks, showing which and when data augmentation methods bring accuracy improvements. Interestingly, we find that data augmentation is not very helpful, most likely because the signal patterns useful to discriminate users are too sensitive to the transformations brought by certain data augmentation techniques. This result is somewhat contradictory to the common belief that data augmentation is expected to increase the accuracy of machine learning models.

Summary (the publisher’s abstract is not openly available): A book on unsupervised visual learning in space and time. It proposes a general unsupervised learning model that brings together graphs and deep neural networks, covering unsupervised graph and hypergraph matching and clustering, feature selection that meets unsupervised learning, object segmentation in video from highly probable positive features, coupling appearance and motion through space and time, and learning over several generations of teacher and student networks — from pixels to seeing. Reported by BookAuthority as the best-selling book on unsupervised learning since 2020.

This paper proposes a real-time Virtual Reality game for treating acrophobia that automatically tailors in-game exposure to heights to the players’ individual characteristics – affective state and physiological features. The elements of novelty are the automatic estimation of fear and the prediction of the next game level based on the electroencephalogram (EEG) and biophysical data – Galvanic Skin Response (GSR) and Heart Rate (HR). Two neural networks have been trained with the data recorded in an experiment where 4 subjects have been in-vivo and virtually exposed to various heights. In order to test the validity of the approach, the same users played the acrophobia game, using two modalities of ex-pressing fear level. After completing a game level, the EEG and biophysical data were averaged and one neural network estimated the current fear score, while the other predicted the next game level. A measure of similarity between the self-estimated fear level during a game epoch and the fear level predicted by the first neural network showed an accuracy rate of 73% and 42% respectively for the two modalities of expressing fear level. 3 out of 4 users succeeded to obtain a fear level of 0 (complete relaxation) in the final game epoch.

When driving, people make decisions based on current traffic as well as their desired route. They have a mental map of known routes and are often able to navigate without needing directions. Current self-driving models improve their performances when using additional GPS information. Here we aim to push forward self-driving research and perform route planning even in the absence of GPS. Our system learns to predict in real-time vehicle’s current location and future trajectory, as a function of time, on a known map, given only the raw video stream and the intended destination. The GPS signal is available only at training time, with training data annotation being fully automatic. Different from other published models, we predict the vehicle’s trajectory for up to seven seconds ahead, from which complete steering, speed and acceleration information can be derived for the entire time span. Trajectories capture navigational information on multiple levels, from instant steering commands that depend on present traffic and obstacles ahead, to longer-term navigation decisions, towards a specific destination. We collect our dataset with a regular car and a smartphone that records video and GPS streams. The GPS data is used to derive ground-truth supervision labels and create an analytical representation of the traversed map. In tests, our system outperforms published methods on visual localization and steering and gives accurate navigation assistance between any two known locations.

Learning in the space-time domain remains a very challenging problem in machine learning and computer vision. Current computational models for understanding spatio-temporal visual data are heavily rooted in the classical single-image based paradigm. It is not yet well understood how to integrate information in space and time into a single, general model. We propose a neural graph model, recurrent in space and time, suitable for capturing both the local appearance and the complex higher-level interactions of different entities and objects within the changing world scene. Nodes and edges in our graph have dedicated neural networks for processing information. Nodes operate over features extracted from local parts in space and time and previous memory states. Edges process messages between connected nodes at different locations and spatial scales or between past and present time. Messages are passed iteratively in order to transmit information globally and establish long range interactions. Our model is general and could learn to recognize a variety of high level spatio-temporal concepts and be applied to different learning tasks. We demonstrate, through extensive experiments and ablation studies, that our model outperforms strong baselines and top published methods on recognizing complex activities in video. Moreover, we obtain state-of-the-art performance on the challenging Something-Something human-object interaction dataset.

The field of affective computing has significantly emerged over the recent years, either we talk about systems that recognize and interpret human emotions or applications with assistive or therapeutic purposes that aim to treat various psychological disorders. Our goal is to design and develop a system that would aid people suffering of a common anxiety disorder of our times - the phobia. Thus, we propose a system for treating acrophobia - fear of heights - that exposes the patients in a Virtual Reality scenario depicting a mountain landscape, while their physiological signals (electrodermal activity, heart rate and respiration rate) are continuously recorded. However, during exposure, certain artefacts can appear, caused by head or hand movements, controller or mouse clicks, body rotations - and thus affect the quality of the recording. As artefact identification is a fundamental step in obtaining clean data, in this paper we present a method of recognizing artefact patterns in the signals recorded during gameplay sessions, which consists of aligning the data segments corresponding to the moments when the users performed head/hand/body movements with the artefact signals recorded during a reference procedure. The results obtained in pilot experiment with 5 users showed that the bias is lower on the perfectly aligned segments than on the segments located a few steps before and after the moment when the user introduced the artefact in the game signal, but without statistical significance. In addition, we propose a new method of estimating respiration rate based on the distance between two HTC Vive trackers placed on an elastic band, on the chest and on the upper back torso of the user.

We propose Shift R-CNN, a hybrid model for monocular 3D object detection, which combines deep learning with the power of geometry. We adapt a Faster R-CNN network for regressing initial 2D and 3D object properties and combine it with a least squares solution for the inverse 2D to 3D geometric mapping problem, using the camera projection matrix. The closed-form solution of the mathematical system, along with the initial output of the adapted Faster R-CNN are then passed through a final ShiftNet network that refines the result using our newly proposed Volume Displacement Loss. Our novel, geometrically constrained deep learning approach to monocular 3D object detection obtains top results on KITTI 3D Object Detection Benchmark, being the best among all monocular methods that do not use any pre-trained network for depth estimation.

We address the challenging task of foreground object discovery and segmentation in video. We introduce an efficient solution, suitable for both unsupervised and supervised scenarios, based on a spacetime graph representation of the video sequence. We ensure a fine grained representation with one-to-one correspondences between graph nodes and video pixels. We formulate the task as a spectral clustering problem by exploiting the spatio-temporal consistency between the scene elements in terms of motion and appearance. Graph nodes that belong to the main object of interest should form a strong cluster, as they are linked through long range optical flow chains and have similar motion and appearance features along those chains. On one hand, the optimization problem aims to maximize the segmentation clustering score based on the motion structure through space and time. On the other hand, the segmentation should be consistent with respect to node features. Our approach leads to a graph formulation in which the segmentation solution becomes the principal eigenvector of a novel Feature-Motion matrix. While the actual matrix is not computed explicitly, the proposed algorithm efficiently computes, in a few iteration steps, the principal eigenvector that captures the segmentation of the main object in the video. The proposed algorithm, GO-VOS, produces a global optimum solution and, consequently, it does not depend on initialization. In practice, GO-VOS achieves state of the art results on three challenging datasets used in current literature: DAVIS, SegTrack and YouTube-Objects.

This paper presents a human-centered methodology for designing and developing Virtual Reality Exposure Therapy (VRET) systems. By following the steps proposed by the methodology – Users analysis, Domain Analysis, Task Analysis and Representational Analysis, we developed a system for acrophobia therapy composed of 9 functional, interrelated modules which are responsible for patients, scenes, audio and graphics management, as well as for physiological monitoring and events triggering. The therapist visualizes in real time the patient’s biophysical signals and adapts the exposure scenario accordingly, as he can lower or increase the level of exposure. There are 3 scenes in the game, containing a ride by cable car, one by ski lift and a walk by foot in a mountain landscape. A reward system is implemented and emotion dimension ratings are collected at predefined points in the scenario. They will be stored and later used for constructing an automatic machine learning emotion recognition and exposure adaptation module.

Semantic segmentation is a crucial task for robot navigation and safety. However, it requires huge amounts of pixelwise annotations to yield accurate results. While recent progress in computer vision algorithms has been heavily boosted by large ground-level datasets, the labeling time has hampered progress in low altitude UAV applications, mostly due to the difficulty imposed by large object scales and pose variations. Motivated by the lack of a large video aerial dataset, we introduce a new one, with high resolution (4K) images and manually-annotated dense labels every 50 frames. To help the video labeling process, we make an important step towards automatic annotation and propose SegProp, an iterative flow-based method with geometric constrains to propagate the semantic labels to frames that lack human annotations. This results in a dataset with more than 50k annotated frames - the largest of its kind, to the best of our knowledge. Our experiments show that SegProp surpasses current state-of-the-art label propagation methods by a significant margin. Furthermore, when training a semantic segmentation deep neural net using the automatically annotated frames, we obtain a compelling overall performance boost at test time of 16.8% mean F-measure over a baseline trained only with manually-labeled frames. Our Ruralscapes dataset, the label propagation code and a fast segmentation tool are available at our website: https://sites.google.com/site/aerialimageunderstanding/

Unsupervised learning poses one of the most difficult challenges in computer vision today. The task has an immense practical value with many applications in artificial intelligence and emerging technologies, as large quantities of unlabeled videos can be collected at relatively low cost. In this paper, we address the unsupervised learning problem in the context of detecting the main foreground objects in single images. We train a student deep network to predict the output of a teacher pathway that performs unsupervised object discovery in videos or large image collections. Our approach is different from published methods on unsupervised object discovery. We move the unsupervised learning phase during training time, then at test time we apply the standard feed-forward processing along the student pathway. This strategy has the benefit of allowing increased generalization possibilities during training, while remaining fast at testing. Our unsupervised learning algorithm can run over several generations of student-teacher training. Thus, a group of student networks trained in the first generation collectively create the teacher at the next generation. In experiments our method achieves top results on three current datasets for object discovery in video, unsupervised image segmentation and saliency detection. At test time the proposed system is fast, being one to two orders of magnitude faster than published unsupervised methods.

Semantic segmentation and vision-based geolocalization in aerial images are challenging tasks in computer vision. Due to the advent of deep convolutional nets and the availability of relatively low cost UAVs, they are currently generating a growing attention in the field. We propose a novel multi-task multi-stage neural network that is able to handle the two problems at the same time, in a single forward pass. The first stage of our network predicts pixelwise class labels, while the second stage provides a precise location using two branches. One branch uses a regression network, while the other is used to predict a location map trained as a segmentation task. From a structural point of view, our architecture uses encoder-decoder modules at each stage, having the same encoder structure re-used. Furthermore, its size is limited to be tractable on an embedded GPU. We achieve commercial GPS-level localization accuracy from satellite images with spatial resolution of 1 square meter per pixel in a city-wide area of interest. On the task of semantic segmentation, we obtain state-of-the-art results on two challenging datasets, the Inria Aerial Image Labeling dataset and Massachusetts Buildings.

Object tracking is an essential problem in computer vision that has been researched for several decades. One of the main challenges in tracking is to adapt to object appearance changes over time and avoiding drifting to background clutter. We address this challenge by proposing a deep neural network composed of different parts, which functions as a society of tracking parts. They work in conjunction according to a certain policy and learn from each other in a robust manner, using co-occurrence constraints that ensure robust inference and learning. From a structural point of view, our network is composed of two main pathways. One pathway is more conservative. It carefully monitors a large set of simple tracker parts learned as linear filters over deep feature activation maps. It assigns the parts different roles. It promotes the reliable ones and removes the inconsistent ones. We learn these filters simultaneously in an efficient way, with a single closed-form formulation, for which we propose novel theoretical properties. The second pathway is more progressive. It is learned completely online and thus it is able to better model object appearance changes. In order to adapt in a robust manner, it is learned only on highly confident frames, which are decided using co-occurrences with the first pathway. Thus, our system has the full benefit of two main approaches in tracking. The larger set of simpler filter parts offers robustness, while the full deep network learned online provides adaptability to change. As shown in the experimental section, our approach achieves state of the art performance on the challenging VOT17 benchmark, outperforming the published methods both on the general EAO metric and in the number of fails, by a significant margin.

Describing visual data into natural language is a very challenging task, at the intersection of computer vision, natural language processing and machine learning. Language goes well beyond the description of physical objects and their interactions and can convey the same abstract idea in many ways. It is both about content at the highest semantic level as well as about fluent form. Here we propose an approach to describe videos in natural language by reaching a consensus among multiple encoder-decoder networks. Finding such a consensual linguistic description, which shares common properties with a larger group, has a better chance to convey the correct meaning. We propose and train several network architectures and use different types of image, audio and video features. Each model produces its own description of the input video and the best one is chosen through an efficient, two-phase consensus process. We demonstrate the strength of our approach by obtaining state of the art results on the challenging MSR-VTT dataset.

Road detection from aerial images is a challenging task for humans and machines alike. Occlusion, the lack of visual cues and slim class borders for other road-like structures (such as pathways or private alleys) make the problem inherently ambiguous, requiring logic that goes beyond the input image. We propose a three-stage method for the task of road segmentation - first, an ensemble of multiple U-Net like CNNs generate binary road masks. Second, another CNN learns to refine roads segmentations based on the fusion of the road maps from the first stage. Third, missing links are added based on the inferred graph to improve segmentation.

The emergence of relatively low cost UAVs has prompted a global concern about the safe operation of such devices. Since most of them can ‘autonomously’ fly by means of GPS way-points, the lack of a higher logic for emergency scenarios leads to an abundance of incidents involving property or personal injury. In order to tackle this problem, we propose a small, embeddable ConvNet for both depth and safe landing area estimation. Furthermore, since labeled training data in the 3D aerial field is scarce and ground images are unsuitable, we capture a novel synthetic aerial 3D dataset obtained from 3D reconstructions. We use the synthetic data to learn to estimate depth from in-flight images and segment them into ‘safe-landing’ and ‘obstacle’ regions. Our experiments demonstrate compelling results in practice on both synthetic data and real RGB drone footage.

Recognizing roads and intersections in aerial images is a challenging problem in computer vision with many real world applications, such as localization and navigation for unmanned aerial vehicles (UAVs). The problem is currently gaining momentum in computer vision and is still far from being solved. While recent approaches have greatly improved due to the advances in deep learning, they provide only pixel-level semantic segmentations. In this paper, we argue that roads and intersections should be recognized at the higher semantic level of road graphs - with roads being edges that connect nodes. Towards this goal we present a method consisting of two stages. During the first stage, we detect roads and intersections with a novel, dual-hop generative adversarial network (DH-GAN) that segments images at the level of pixels. At the second stage, given the pixelwise road segmentation, we find its best covering road graph by applying a smoothing-based graph optimization procedure. Our approach is able to outperform recent published methods and baselines on a large dataset with European roads.

The importance of visual context in object recognition has been intensively studied over the years. Along with the advent of deep convolutional neural networks (CNN), using contextual information with such systems starts to receive attention in the literature. Regardless of deep learning advances, aerial image analysis still poses many great challenges. Satellite images are often taken under poor lighting conditions and contain low resolution objects, many times occluded. For this particular task, visual context could be of great help, but there are still very few papers that consider context in aerial image understanding.Our work addresses the task of object segmentation in aerial images with a novel dual-stream deep convolutional neural network that integrates the local object appearance and global contextual information into a unified network. Our model learns to combine local object appearance and global semantic knowledge simultaneously and in a complementary way, so that together they form a powerful classifier. Experiments on the Massachusetts Buildings Dataset demonstrate the superiority of our model over state-of-the-art methods. We also introduce two new challenging datasets for the task of buildings and road segmentation. While our local-global model could also be useful in general recognition tasks, we clearly demonstrate the effectiveness of visual context in conjunction with deep nets in aerial image understanding.

We address an essential problem in computer vision, that of unsupervised object segmentation in video, where a main object of interest in a video sequence should be automatically separated from its background. An efficient solution to this task would enable large-scale video interpretation at a high semantic level in the absence of the costly manually labeled ground truth. We propose an efficient unsupervised method for generating foreground object soft-segmentation masks based on automatic selection and learning from highly probable positive features. We show that such features can be selected efficiently by taking into consideration the spatio-temporal, appearance and motion consistency of the object during the whole observed sequence. We also emphasize the role of the contrasting properties between the foreground object and its background. Our model is created in two stages: we start from pixel level analysis, on top of which we add a regression model trained on a descriptor that considers information over groups of pixels and is both discriminative and invariant to many changes that the object undergoes throughout the video. We also present theoretical properties of our unsupervised learning method, that under some mild constraints is guaranteed to learn a correct discriminative classifier even in the unsupervised case. Our method achieves competitive and even state of the art results on the challenging Youtube-Objects and SegTrack datasets, while being at least one order of magnitude faster than the competition. We believe that the competitive performance of our method in practice, along with its theoretical properties, constitute an important step towards solving unsupervised discovery in video.

Aerial images are often taken under poor lighting conditions and contain low resolution objects, many times occluded by other objects. In this domain, visual context could be of great help, but there are still very few papers that consider context in aerial image understanding and still remains an open problem in computer vision. We propose a dual-stream deep neural network that processes information along two independent pathways. Our model learns to combine local and global appearance in a complementary way, such that together form a powerful classifier. We test our dual-stream network on the task of buildings segmentation in aerial images and obtain state-of-the-art results on the Massachusetts Buildings Dataset. We study the relative importance of local appearance versus the larger scene, as well as their performance in combination on three new buildings datasets. We clearly demonstrate the effectiveness of visual context in conjunction with deep neural networks for aerial image understanding.

Aerial image analysis at a semantic level is important in many applications with strong potential impact in industry and consumer use, such as automated mapping, urban planning, real estate and environment monitoring, or disaster relief. The problem is enjoying a great interest in computer vision and remote sensing, due to increased computer power and improvement in automated image understanding algorithms. In this paper we address the task of automatic geolocalization of aerial images from recognition and matching of roads and intersections. Our proposed method is a novel contribution in the literature that could enable many applications of aerial image analysis when GPS data is not available. We offer a complete pipeline for geolocalization, from the detection of roads and intersections, to the identification of the enclosing geographic region by matching detected intersections to previously learned manually labeled ones, followed by accurate geometric alignment between the detected roads and the manually labeled maps. We test on a novel dataset with aerial images of two European cities and use the publicly available OpenStreetMap project for collecting ground truth roads annotations. We show in extensive experiments that our approach produces highly accurate localizations in the challenging case when we train on images from one city and test on the other and the quality of the aerial images is relatively poor. We also show that the the alignment between detected roads and pre-stored manual annotations can be effectively used for improving the quality of the road detection results.

Visual context is important in object recognition and it is still an open problem in computer vision. Along with the advent of deep convolutional neural networks (CNN), using contextual information with such systems starts to receive attention in the literature. At the same time, aerial imagery is gaining momentum. While advances in deep learning make good progress in aerial image analysis, this problem still poses many great challenges. Aerial images are often taken under poor lighting conditions and contain low resolution objects, many times occluded by trees or taller buildings. In this domain, in particular, visual context could be of great help, but there are still very few papers that consider context in aerial image understanding. Here we introduce context as a complementary way of recognizing objects. We propose a dual-stream deep neural network model that processes information along two independent pathways, one for local and another for global visual reasoning. The two are later combined in the final layers of processing. Our model learns to combine local object appearance as well as information from the larger scene at the same time and in a complementary way, such that together they form a powerful classifier. We test our dual-stream network on the task of segmentation of buildings and roads in aerial images and obtain state-of-the-art results on the Massachusetts Buildings Dataset. We also introduce two new datasets, for buildings and road segmentation, respectively, and study the relative importance of local appearance vs. the larger scene, as well as their performance in combination. While our local-global model could also be useful in general recognition tasks, we clearly demonstrate the effectiveness of visual context in conjunction with deep nets for aerial image understanding.