Human motion generation has shown great advances powered by recent diffusion models and large-scale motion capture data. Most of existing works, however, currently target animation of isolated people in empty scenes. Meanwhile, synthesizing realistic human-object interactions in complex 3D scenes remains a critical challenge in computer graphics and robotics. One obstacle towards generating versatile high-fidelity humanobject interactions is the lack of large-scale datasets with diverse object manipulations. Indeed, existing motion capture data is typically restricted to single people and manipulations of limited sets of objects. To address this issue, we propose an automatic motion extraction pipeline and use it to collect interaction-rich human motions. Our new dataset InterPose contains 73.8 K sequences of 3D human motions and corresponding text captions automatically obtained from 45.8 K videos with human-object interactions. We perform extensive experiments and demonstrate InterPose to bring significant improvements to state-of-the-art methods for human motion generation. Moreover, using InterPose we develop an LLM-based agent enabling zero-shot animation of people interacting with diverse objects and scenes.

Vision-And-Language Navigation (VLN) suffers from the limited diversity and scale of training data, primarily constrained by the manual curation of existing simulators. To address this, we introduce RoomTour3D, a video-instruction dataset derived from web-based room tour videos that capture real-world indoor spaces and human walking demonstrations. Unlike existing VLN datasets, RoomTour3D leverages the scale and diversity of online videos to generate open-ended human walking trajectories and open-world navigable instructions. To compensate for the lack of navigation data in online videos, we perform 3D reconstruction and obtain 3D trajectories of walking paths augmented with additional information on the room types, object locations and 3D shape of surrounding scenes. Our dataset includes ~100K open-ended description-enriched trajectories with ~200K instructions, and 17K action-enriched trajectories from 1847 room tour environments. We demonstrate experimentally that RoomTour3D enables significant improvements across multiple VLN tasks including CVDN, SOON, R2R, and REVERIE. Moreover, Room-Tour3D facilitates the development of trainable zero-shot VLN agents, showcasing the potential and challenges of advancing towards open-world navigation.

The goal of this work is to generate step-by-step visual instructions in the form of a sequence of images, given an input image that provides the scene context and the sequence of textual instructions. This is a challenging problem as it requires generating multi-step image sequences to achieve a complex goal while being grounded in a specific environment. Part of the challenge stems from the lack of large-scale training data for this problem. The contribution of this work is thus three-fold. First, we introduce an automatic approach for collecting large step-by-step visual instruction training data from instructional videos. We apply this approach to one million videos and create a large-scale, high-quality dataset of 0.6M sequences of image-text pairs. Second, we develop and train ShowHowTo, a video diffusion model capable of generating step-by-step visual instructions consistent with the provided input image. Third, we evaluate the generated image sequences across three dimensions of accuracy (step, scene, and task) and show our model achieves state-of-the-art results on all of them. Our code, dataset, and trained models are publicly available.

Existing Large Multimodal Models (LMMs) generally focus on only a few regions and languages. As LMMs continue to improve, it is increasingly important to ensure they understand cultural contexts, respect local sensitivities, and support low-resource languages, all while effectively integrating corresponding visual cues. In pursuit of culturally diverse global multimodal models, our proposed All Languages Matter Benchmark (ALM-bench) represents the largest and most comprehensive effort to date for evaluating LMMs across 100 languages. ALM-bench challenges existing models by testing their ability to understand and reason about culturally diverse images paired with text in various languages, including many low-resource languages traditionally underrepresented in LMM research. The benchmark offers a robust and nuanced evaluation framework featuring various question formats, including true/false, multiple choice, and open-ended questions, which are further divided into short and long-answer categories. ALM-bench design ensures a comprehensive assessment of a model’s ability to handle varied levels of difficulty in visual and linguistic reasoning. To capture the rich tapestry of global cultures, ALM-bench carefully curates content from 13 distinct cultural aspects, ranging from traditions and rituals to famous personalities and celebrations. Through this, ALM-bench not only provides a rigorous testing ground for state-of-the-art open and closed-source LMMs but also highlights the importance of cultural and linguistic inclusivity, encouraging the development of models that can serve diverse global populations effectively. Our benchmark is publicly available at https://mbzuai-oryx.github.io/ALM-Bench/.

Bhuiyan Sanjid Shafique, Ashmal Vayani, Muhammad Maaz, Hanoona Abdul Rasheed, Dinura Dissanayake, Mohammed Irfan Kurpath, Yahya Hmaiti, Go Inoue, Jean Lahoud, Md. Safirur Rashid, Shadid Intisar Quasem, Maheen Fatima, Franco Vidal, Mykola Maslych, Ketan Pravin More, Sanoojan Baliah, Hasindri Watawana, Yuhao Li, Fabian Farestam, Leon Schaller, Roman Tymtsiv, Simon Weber, Hisham Cholakkal, Ivan Laptev, Shin’ichi Satoh, Michael Felsberg, Mubarak Shah, Salman Khan, Fahad Shahbaz Khan. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

In this work, we aim to learn a unified vision-based policy for multi-fingered robot hands to manipulate a variety of objects in diverse poses. Though prior work has shown benefits of using human videos for policy learning, performance gains have been limited by the noise in estimated trajectories. Moreover, reliance on privileged object information such as ground-truth object states further limits the applicability in realistic scenarios. To address these limitations, we propose a new framework ViViDex to improve vision-based policy learning from human videos. It first uses reinforcement learning with trajectory guided rewards to train state-based policies for each video, obtaining both visually natural and physically plausible trajectories from the video. We then rollout successful episodes from state-based policies and train a unified visual policy without using any privileged information. We propose coordinate transformation to further enhance the visual point cloud representation, and compare behavior cloning and diffusion policy for the visual policy training. Experiments both in simulation and on the real robot demonstrate that ViViDex outperforms state-of-theart approaches on three dexterous manipulation tasks. Project website: zerchen.github.io/projects/vividex.html.

While large multimodal models (LMMs) have demonstrated strong performance across various Visual Question Answering (VQA) tasks, certain challenges require complex multi-step reasoning to reach accurate answers. One particularly challenging task is autonomous driving, which demands thorough cognitive processing before decisions can be made. In this domain, a sequential and interpretive understanding of visual cues is essential for effective perception, prediction, and planning. Nevertheless, common VQA benchmarks often focus on the accuracy of the final answer while overlooking the reasoning process that enables the generation of accurate responses. Moreover, existing methods lack a comprehensive framework for evaluating step-by-step reasoning in realistic driving scenarios. To address this gap, we propose DriveLMM-o1, a new dataset and benchmark specifically designed to advance step-wise visual reasoning for autonomous driving. Our benchmark features over 18k VQA examples in the training set and more than 4k in the test set, covering diverse questions on perception, prediction, and planning, each enriched with step-by-step reasoning to ensure logical inference in autonomous driving scenarios. We further introduce a large multimodal model that is fine-tuned on our reasoning dataset, demonstrating robust performance in complex driving scenarios. In addition, we benchmark various open-source and closed-source methods on our proposed dataset, systematically comparing their reasoning capabilities for autonomous driving tasks. Our model achieves a +7.49% gain in final answer accuracy, along with a 3.62% improvement in reasoning score over the previous best open-source model. Our framework, dataset, and model are available at https://github.com/ayesha-ishaq/DriveLMM-o1.

Large Language Models (LLMs) have demonstrated remarkable planning abilities across various domains, including robotic manipulation and navigation. While recent work in robotics deploys LLMs for high-level and low-level planning, existing methods often face challenges with failure recovery and suffer from hallucinations in long-horizon tasks. To address these limitations, we propose a novel multi-agent LLM framework, Multi-Agent Large Language Model for Manipulation (MALMM). Notably, MALMM distributes planning across three specialized LLM agents, namely high-level planning agent, low-level control agent, and a supervisor agent. Moreover, by incorporating environment observations after each step, our framework effectively handles intermediate failures and enables adaptive re-planning. Unlike existing methods, MALMM does not rely on pre-trained skill policies or in-context learning examples and generalizes to unseen tasks. In our experiments, MALMM demonstrates excellent performance in solving previously unseen long-horizon manipulation tasks, and outperforms existing zero-shot LLM-based methods in RLBench by a large margin. Experiments with the Franka robot arm further validate our approach in real-world settings.

Changing facial expressions, gestures, or background details may dramatically alter the meaning conveyed by an image. Notably, recent advances in diffusion models greatly improve the quality of image manipulation while also opening the door to misuse. Identifying changes made to authentic images, thus, becomes an important task, constantly challenged by new diffusion-based editing tools. To this end, we propose a novel approach for ReliAble iDentification of inpainted AReas (RADAR). RADAR builds on existing foundation models and combines features from different image modalities. It also incorporates an auxiliary contrastive loss that helps to isolate manipulated image patches. We demonstrate these techniques to significantly improve both the accuracy of our method and its generalisation to a large number of diffusion models. To support realistic evaluation, we further introduce BBC-PAIR, a new comprehensive benchmark, with images tampered by 28 diffusion models. Our experiments show that RADAR achieves excellent results, outperforming the state-of-the-art in detecting and localising image edits made by both seen and unseen diffusion models. Further information about our code, data and models, including separate licensing terms, will be publicly available at https://alex-costanzino.github.io/radar/.

Efficient fine-tuning of pre-trained Text-to-Image (T2I) models involves adjusting the model to suit a particular task or dataset while minimizing computational resources and limiting the number of trainable parameters. However, it often faces challenges in striking a trade-off between aligning with the target distribution: learning a novel concept from a limited image for personalization and retaining the instruction ability needed for unifying multiple tasks, all while maintaining editability (aligning with a variety of prompts or in-context generation). In this work, we introduce DEFT, Decompositional Efficient Fine-Tuning, an efficient fine-tuning framework that adapts a pre-trained weight matrix by decomposing its update into two components with two trainable matrices: (1) a projection onto the complement of a low-rank subspace spanned by a low-rank matrix, and (2) a low-rank update. The single trainable low-rank matrix defines the subspace, while the other trainable low-rank matrix enables parameter adaptation within that subspace. We conducted extensive experiments on the Dreambooth and Dreambench Plus datasets for personalization, the InsDet dataset for object and scene adaptation, and the VisualCloze dataset for a universal image generation framework through visual in-context learning with both Stable Diffusion and a unified model. Our results demonstrated state-of-the-art performance, highlighting the emergent properties of efficient fine-tuning. Our code is available on \href{https://github.com/MAXNORM8650/DEFT}{DEFT}.

While vision-language models (VLMs) have demonstrated promising capabilities in reasoning and planning for embodied agents, their ability to comprehend physical phenomena, particularly within structured 3D environments, remains severely limited. To close this gap, we introduce PhyBlock, a progressive benchmark designed to assess VLMs on physical understanding and planning through robotic 3D block assembly tasks. PhyBlock integrates a novel four-level cognitive hierarchy assembly task alongside targeted Visual Question Answering (VQA) samples, collectively aimed at evaluating progressive spatial reasoning and fundamental physical comprehension, including object properties, spatial relationships, and holistic scene understanding. PhyBlock includes 2600 block tasks (400 assembly tasks, 2200 VQA tasks) and evaluates models across three key dimensions: partial completion, failure diagnosis, and planning robustness. We benchmark 23 state-of-the-art VLMs, highlighting their strengths and limitations in physically grounded, multi-step planning. Our empirical findings indicate that the performance of VLMs exhibits pronounced limitations in high-level planning and reasoning capabilities, leading to a notable decline in performance for the growing complexity of the tasks.Error analysis reveals persistent difficulties in spatial orientation and dependency reasoning.We position PhyBlock as a unified testbed to advance embodied reasoning, bridging vision-language understanding and real-world physical problem-solving.

Recent methods for visual question answering rely on large-scale annotated datasets. Manual annotation of questions and answers for videos, however, is tedious, expensive and prevents scalability. In this work, we propose to avoid manual annotation and generate a large-scale training dataset for video question answering making use of automatic cross-modal supervision. We leverage a question generation transformer trained on text data and use it to generate question-answer pairs from transcribed video narrations. Given narrated videos, we then automatically generate the HowToVQA69M dataset with 69M video-question-answer triplets. To handle the open vocabulary of diverse answers in this dataset, we propose a training procedure based on a contrastive loss between a video-question multi-modal transformer and an answer transformer. We introduce the zero-shot VideoQA task and the VideoQA feature probe evaluation setting and show excellent results, in particular for rare answers. Furthermore, our method achieves competitive results on MSRVTT-QA, ActivityNet-QA, MSVD-QA and How2QA datasets. We also show that our VideoQA dataset generation approach generalizes to another source of web video and text data. We use our method to generate the WebVidVQA3M dataset from the WebVid dataset, i.e., videos with alt-text annotations, and show its benefits for training VideoQA models. Finally, for a detailed evaluation we introduce iVQA, a new VideoQA dataset with reduced language bias and high-quality manual annotations. Code, datasets and trained models are available on our project webpage (https://antoyang.github.io/just-ask.html).

Image and video analysis requires not only accurate object detection but also the understanding of relationships among detected objects. Common solutions to relation modeling typically resort to stand-alone object detectors followed by non-differentiable post-processing techniques. Re-cently introduced detection transformers (DETR) perform end-to-end object detection based on a bipartite matching loss. Such methods, however, lack the ability to jointly detect objects and resolve object associations. In this paper, we build on the DETR approach and extend it to the joint detection of objects and their relationships by introducing an approximated bipartite matching. While our method can generalize to an arbitrary number of objects, we here focus on the modeling of object pairs and their relations. In particular, we apply our method PairDETR to the problem of detecting human bodies and faces, and associating them for the same person. Our approach not only eliminates the need for hand-designed post-processing but also achieves excellent results for body-face associations. We evaluate PairDETR on the challenging CrowdHuman and CityPersons datasets and demonstrate a large improvement over the state of the art. Our training code and pretrained models are available at https://github.com/mts-ai/pairdetr

Learning generalizable visual representations from Internet data has yielded promising results for robotics. Yet, prevailing approaches focus on pre-training 2D representations, being sub-optimal to deal with occlusions and accurately localize objects in complex 3D scenes. Meanwhile, 3D representation learning has been limited to single-object understanding. To address these limitations, we introduce a novel 3D pre-training framework for robotics named SUGAR that captures semantic, geometric and affordance properties of objects through 3D point clouds. We underscore the importance of cluttered scenes in 3D representation learning, and automatically construct a multi-object dataset benefiting from cost-free supervision in simulation. SUGAR employs a versatile transformer-based model to jointly address five pre-training tasks, namely cross-modal knowledge distillation for semantic learning, masked point modeling to understand geometry structures, grasping pose synthesis for object affordance, 3D instance segmentation and referring expression grounding to analyze cluttered scenes. We evaluate our learned representation on three robotic-related tasks, namely, zero-shot 3D object recognition, referring expression grounding, and language-driven robotic manipulation. Experimental results show that SUGAR's 3D representation outperforms state-of-the-art 2D and 3D representations.

We address the task of generating temporally consistent and physically plausible images of actions and object state transformations. Given an input image and a text prompt describing the targeted transformation, our generated images preserve the environment and transform objects in the initial image. Our contributions are threefold. First, we leverage a large body of instructional videos and automatically mine a dataset of triplets of consecutive frames corresponding to initial object states, actions, and resulting object transformations. Second, equipped with this data, we develop and train a conditioned diffusion model dubbed GenHowTo. Third, we evaluate GenHowTo on a variety of objects and actions and show superior performance compared to existing methods. In particular, we introduce a quantitative evaluation where GenHowTo achieves 88% and 74% on seen and unseen interaction categories, respectively, outperforming prior work by a large margin.

Recent Large Vision Language Models (LVLMs) present remarkable zero-shot conversational and reasoning capabilities given multimodal queries. Nevertheless, they suffer from object hallucination, a phenomenon where LVLMs are prone to generate textual responses not factually aligned with image inputs. Our pilot study reveals that object hallucination is closely tied with Rotary Position Encoding (RoPE), a widely adopted positional dependency modeling design in existing LVLMs. Due to the long-term decay in RoPE, LVLMs tend to hallucinate more when relevant visual cues are distant from instruction tokens in the multimodal input sequence, Additionally, we observe a similar effect when reversing the sequential order of visual tokens during multimodal alignment. Our tests indicate that long-term decay in RoPE poses challenges to LVLMs while capturing visual-instruction interactions across long distances. We propose Concentric Causal Attention (CCA), a simple yet effective positional alignment strategy that mitigates the impact of RoPE long-term decay in LVLMs by naturally reducing relative distance between visual and instruction tokens. With CCA, visual tokens can better interact with instruction tokens, thereby enhancing model's perception capability and alleviating object hallucination. Without bells and whistles, our positional alignment method surpasses existing hallucination mitigation strategies by large margins on multiple object hallucination benchmarks.

We aim to learn to temporally localize object state changes and the corresponding state-modifying actions by observing people interacting with objects in long uncurated web videos. We introduce three principal contributions. First, we develop a self-supervised model for jointly learning state-modifying actions together with the corresponding object states from an uncurated set of videos from the Internet. The model is self-supervised by the causal ordering signal, i.e., initial object state → manipulating action → end state. Second, we explore alternative multi-task network architectures and identify a model that enables efficient joint learning of multiple object states and actions, such as pouring water and pouring coffee, together. Third, we collect a new dataset, named ChangeIt, with more than 2600 hours of video and 34 thousand changes of object states. We report results on an existing instructional video dataset COIN as well as our new large-scale ChangeIt dataset containing tens of thousands of long uncurated web videos depicting various interactions such as hole drilling, cream whisking, or paper plane folding. We show that our multi-task model achieves a relative improvement of 40% over the prior methods and significantly outperforms both image-based and video-based zero-shot models for this problem.

Physics simulation is ubiquitous in robotics. Whether in model-based approaches (e.g., trajectory optimization), or model-free algorithms (e.g., reinforcement learning), physics simulators are a central component of modern control pipelines in robotics. Over the past decades, several robotic simulators have been developed, each with dedicated contact modeling assumptions and algorithmic solutions. In this article, we survey the main contact models and the associated numerical methods commonly used in robotics for simulating advanced robot motions involving contact interactions. In particular, we recall the physical laws underlying contacts and friction (i.e., Signorini condition, Coulomb's law, and the maximum dissipation principle), and how they are transcribed in current simulators. For each physics engine, we expose their inherent physical relaxations along with their limitations due to the numerical techniques employed. Based on our study, we propose theoretically grounded quantitative criteria on which we build benchmarks assessing both the physical and computational aspects of simulation. We support our work with an open-source and efficient C++ implementation of the existing algorithmic variations. Our results demonstrate that some approximations or algorithms commonly used in robotics can severely widen the reality gap and impact target applications. We hope this work will help motivate the development of new contact models, contact solvers, and robotic simulators in general, at the root of recent progress in motion generation in robotics.

One of the major challenges of machine translation (MT) is ambiguity, which can in some cases be resolved by accompanying context such as images.However, recent work in multimodal MT (MMT) has shown that obtaining improvements from images is challenging, limited not only by the difficulty of building effective cross-modal representations, but also by the lack of specific evaluation and training data.We present a new MMT approach based on a strong text-only MT model, which uses neural adapters, a novel guided self-attention mechanism and which is jointly trained on both visually-conditioned masking and MMT.We also introduce CoMMuTE, a Contrastive Multilingual Multimodal Translation Evaluation set of ambiguous sentences and their possible translations, accompanied by disambiguating images corresponding to each translation.Our approach obtains competitive results compared to strong text-only models on standard English→French, English→German and English→Czech benchmarks and outperforms baselines and state-of-the-art MMT systems by a large margin on our contrastive test set.Our code 1 and CoMMuTE 2 are freely available.8 mBART is pretrained on CC25 (Wenzek et al., 2020).

Signed distance functions (SDFs) is an attractive frame-work that has recently shown promising results for 3D shape reconstruction from images. SDFs seamlessly generalize to different shape resolutions and topologies but lack explicit modelling of the underlying 3D geometry. In this work, we ex-ploit the hand structure and use it as guidance for SDF-based shape reconstruction. In particular, we address reconstruction of hands and manipulated objects from monocular RGB images. To this end, we estimate poses of hands and objects and use them to guide 3D reconstruction. More specifically, we predict kinematic chains of pose transformations and align SDFs with highly-articulated hand poses. We improve the visual features of 3D points with geometry alignment and further leverage temporal information to enhance the robustness to occlusion and motion blurs. We conduct extensive experiments on the challenging ObMan and DexYCB benchmarks and demonstrate significant improvements of the proposed method over the state of the art.

In this work, we introduce Vid2Seq, a multi-modal single-stage dense event captioning model pretrained on narrated videos which are readily-available at scale. The Vid2Seq architecture augments a language model with special time tokens, allowing it to seamlessly predict event boundaries and textual descriptions in the same output sequence. Such a unified model requires large-scale training data, which is not available in current annotated datasets. We show that it is possible to leverage unlabeled narrated videos for dense video captioning, by reformulating sentence boundaries of transcribed speech as pseudo event boundaries, and using the transcribed speech sentences as pseudo event captions. The resulting Vid2Seq model pretrained on the YT-Temporal-1B dataset improves the state of the art on a variety of dense video captioning benchmarks including YouCook2, ViTT and ActivityNet Captions. Vid2Seq also generalizes well to the tasks of video paragraph captioning and video clip captioning, and to few-shot settings. Our code is publicly available at [1].

The ability to specify robot commands by a non-expert user is critical for building generalist agents capable of solving a large variety of tasks. One convenient way to specify the intended robot goal is by a video of a person demonstrating the target task. While prior work typically aims to imitate human demonstrations performed in robot environments, here we focus on a more realistic and challenging setup with demonstrations recorded in natural and diverse human environments. We propose Video-conditioned Policy learning (ViP), a data-driven approach that maps human demonstrations of previously unseen tasks to robot manipulation skills. To this end, we learn our policy to generate appropriate actions given current scene observations and a video of the target task. To encourage generalization to new tasks, we avoid particular tasks during training and learn our policy from unlabelled robot trajectories and corresponding robot videos. Both robot and human videos in our framework are represented by video embeddings pre-trained for human action recognition. At test time we first translate human videos to robot videos in the common video embedding space, and then use resulting embeddings to condition our policies. Notably, our approach enables robot control by human demonstrations in a zero-shot manner, i.e., without using robot trajectories paired with human instructions during training. We validate our approach on a set of challenging multi-task robot manipulation environments and outperform state of the art. Our method also demonstrates excellent performance in a new challenging zero-shot setup where no paired data is used during training.

Reinforcement learning (RL) and trajectory opti-mization (TO) present strong complementary advantages. On one hand, RL approaches are able to learn global control policies directly from data, but generally require large sample sizes to properly converge towards feasible policies. On the other hand, TO methods are able to exploit gradient-based information extracted from simulators to quickly converge towards a locally optimal control trajectory which is only valid within the vicinity of the solution. Over the past decade, several approaches have aimed to adequately combine the two classes of methods in order to obtain the best of both worlds. Following on from this line of research, we propose several improvements on top of these approaches to learn global control policies quicker, notably by leveraging sensitivity information stemming from TO methods via Sobolev learning, and Augmented Lagrangian (AL) techniques to enforce the consensus between TO and policy learning. We evaluate the benefits of these improvements on various classical tasks in robotics through comparison with existing approaches in the literature.

Object goal navigation aims to navigate an agent to locations of a given object category in unseen environments. Classical methods explicitly build maps of environments and require extensive engineering while lacking semantic information for object-oriented exploration. On the other hand, end-to-end learning methods alleviate manual map design and predict actions using implicit representations. Such methods, however, lack an explicit notion of geometry and may have limited ability to encode navigation history. In this work, we propose an implicit spatial map for object goal navigation. Our implicit map is recursively updated with new observations at each step using a transformer. To encourage spatial reasoning, we introduce auxiliary tasks and train our model to reconstruct explicit maps as well as to predict visual features, semantic labels and actions. Our method significantly outperforms the state of the art on the challenging MP3D dataset and generalizes well to the HM3D dataset. We successfully deploy our model on a real robot and achieve encouraging object goal navigation results in real scenes using only a few real-world demonstrations. Code, trained models and videos are available at https://www.di.ens.fr/willow/research/onav_rim/.

Learning visuomotor policies in simulation is much safer and cheaper than in the real world. However, due to discrepancies between the simulated and real data, simulator-trained policies often fail when transferred to real robots. One common approach to bridge the visual sim-to-real domain gap is domain randomization (DR). While previous work mainly evaluates DR for disembodied tasks, such as pose estimation and object detection, here we systematically explore visual domain randomization methods and benchmark them on a rich set of challenging robotic manipulation tasks. In particular, we propose an offline proxy task of cube localization to select DR parameters for texture randomization, lighting randomization, variations of object colors and camera parameters. Notably, we demonstrate that DR parameters have similar impact on our offline proxy task and online policies. We, hence, use offline optimized DR parameters to train visuomotor policies in simulation and directly apply such policies to a real robot. Our approach achieves 93% success rate on average when tested on a diverse set of challenging manipulation tasks. Moreover, we evaluate the robustness of policies to visual variations in real scenes and show that our simulator-trained policies outperform policies learned using real but limited data. Code, simulation environment, real robot datasets and trained models are available at https://www.di.ens.fr/willow/research/robust_s2r/.

Segmenting untrimmed videos into chapters enables users to quickly navigate to the information of their interest. This important topic has been understudied due to the lack of publicly released datasets. To address this issue, we present VidChapters-7M, a dataset of 817K user-chaptered videos including 7M chapters in total. VidChapters-7M is automatically created from videos online in a scalable manner by scraping user-annotated chapters and hence without any additional manual annotation. We introduce the following three tasks based on this data. First, the video chapter generation task consists of temporally segmenting the video and generating a chapter title for each segment. To further dissect the problem, we also define two variants of this task: video chapter generation given ground-truth boundaries, which requires generating a chapter title given an annotated video segment, and video chapter grounding, which requires temporally localizing a chapter given its annotated title. We benchmark both simple baselines as well as state-of-the-art video-language models on these three tasks. We also show that pretraining on VidChapters-7M transfers well to dense video captioning tasks, largely improving the state of the art on the YouCook2 and ViTT benchmarks. Finally, our experiments reveal that downstream performance scales well with the size of the pretraining dataset.

Following language instructions to navigate in unseenaenvironments is a challenging problem for autonomous embodied agents. The agent not only needs to ground languages in visual scenes, but also should explore the environment to reach its target. In this work, we propose a dual-scale graph transformer (DUET) for joint long-term action planning and fine-grained cross-modal understanding. We build a topological map on-the-fly to enable efficient exploration in global action space. To balance the complexity of large action space reasoning and fine-grained language grounding, we dynamically combine a fine-scale encoding over local observations and a coarse-scale encoding on a global map via graph transformers. The proposed approach, DUET, significantly outperforms state-of-the-art methods on goal-oriented vision-and-language navigation (VLN) benchmarks REVERIE and SOON. It also improves the success rate on the fine-grained VLN benchmark R2R.

Human actions often induce changes of object states such as “cutting an apple”, “cleaning shoes” or “pouring coffee”. In this paper, we seek to temporally localize object states (e.g. “empty” and “full” cup) together with the corresponding state-modifying actions (“pouring coffee”) in long uncurated videos with minimal supervision. The contributions of this work are threefold. First, we develop a self-supervised model for jointly learning state-modifying actions together with the corresponding object states from an uncurated set of videos from the Internet. The model is self-supervised by the causal ordering signal, i.e. initialobject state → manipulating action → end state. Second, to cope with noisy uncurated training data, our model incorporates a noise adaptive weighting module supervised by a small number of annotated still images, that allows to efficiently filter out irrelevant videos during training. Third, we collect a new dataset with more than 2600 hours of video and 34 thousand changes of object states, and manually annotate a part of this data to validate our approach. Our results demonstrate substantial improvements over prior work in both action and object state-recognition in video.

We consider the problem of localizing a spatio-temporal tube in a video corresponding to a given text query. This is a challenging task that requires the joint and efficient modeling of temporal, spatial and multi-modal interactions. To address this task, we propose TubeDETR, a transformer-based architecture inspired by the recent success of such models for text-conditioned object detection. Our model notably includes: (i) an efficient video and text encoder that models spatial multi-modal interactions over sparsely sampled frames and (ii) a space-time decoder that jointly performs spatio-temporal localization. We demonstrate the advantage of our proposed components through an extensive ablation study. We also evaluate our full approach on the spatio-temporal video grounding task and demonstrate improvements over the state of the art on the challenging VidSTG and HC-STVG benchmarks.

Localizing objects in 3D scenes based on natural language requires understanding and reasoning about spatial relations. In particular, it is often crucial to distinguish similar objects referred by the text, such as "the left most chair" and "a chair next to the window". In this work we propose a language-conditioned transformer model for grounding 3D objects and their spatial relations. To this end, we design a spatial self-attention layer that accounts for relative distances and orientations between objects in input 3D point clouds. Training such a layer with visual and language inputs enables to disambiguate spatial relations and to localize objects referred by the text. To facilitate the cross-modal learning of relations, we further propose a teacher-student approach where the teacher model is first trained using ground-truth object labels, and then helps to train a student model using point cloud inputs. We perform ablation studies showing advantages of our approach. We also demonstrate our model to significantly outperform the state of the art on the challenging Nr3D, Sr3D and ScanRefer 3D object grounding datasets.

Video question answering (VideoQA) is a complex task that requires diverse multi-modal data for training. Manual annotation of question and answers for videos, however, is tedious and prohibits scalability. To tackle this problem, recent methods consider zero-shot settings with no manual annotation of visual question-answer. In particular, a promising approach adapts frozen autoregressive language models pretrained on Web-scale text-only data to multi-modal inputs. In contrast, we here build on frozen bidirectional language models (BiLM) and show that such an approach provides a stronger and cheaper alternative for zero-shot VideoQA. In particular, (i) we combine visual inputs with the frozen BiLM using light trainable modules, (ii) we train such modules using Web-scraped multi-modal data, and finally (iii) we perform zero-shot VideoQA inference through masked language modeling, where the masked text is the answer to a given question. Our proposed approach, FrozenBiLM, outperforms the state of the art in zero-shot VideoQA by a significant margin on a variety of datasets, including LSMDC-FiB, iVQA, MSRVTT-QA, MSVD-QA, ActivityNet-QA, TGIF-FrameQA, How2QA and TVQA. It also demonstrates competitive performance in the few-shot and fully-supervised setting. Our code and models are publicly available at https://github.com/antoyang/FrozenBiLM.

Our work aims to obtain 3D reconstruction of hands and manipulated objects from monocular videos. Reconstructing hand-object manipulations holds a great potential for robotics and learning from human demonstrations. The supervised learning approach to this problem, however, requires 3D supervision and remains limited to constrained laboratory settings and simulators for which 3D ground truth is available. In this paper we first propose a learning-free fitting approach for hand-object reconstruction which can seamlessly handle two-hand object interactions. Our method relies on cues obtained with common methods for object detection, hand pose estimation and instance segmentation. We quantitatively evaluate our approach and show that it can be applied to datasets with varying levels of difficulty for which training data is unavailable.

Our objective is language-based search of large-scale image and video datasets. For this task, the approach that consists of independently mapping text and vision to a joint embedding space, a.k.a. dual encoders, is attractive as retrieval scales and is efficient for billions of images using approximate nearest neighbour search. An alternative approach of using vision-text transformers with cross-attention gives considerable improvements in accuracy over the joint embeddings, but is often inapplicable in practice for large-scale retrieval given the cost of the cross-attention mechanisms required for each sample at test time. This work combines the best of both worlds. We make the following three contributions. First, we equip transformer-based models with a new fine-grained cross-attention architecture, providing significant improvements in retrieval accuracy whilst preserving scalability. Second, we introduce a generic approach for combining a Fast dual encoder model with our Slow but accurate transformer-based model via distillation and reranking. Finally, we validate our approach on the Flickr30K image dataset where we show an increase in inference speed by several orders of magnitude while having results competitive to the state of the art. We also extend our method to the video domain, improving the state of the art on the VATEX dataset.

Vision-and-language navigation (VLN) aims to enable embodied agents to navigate in realistic environments using natural language instructions. Given the scarcity of domain-specific training data and the high diversity of image and language inputs, the generalization of VLN agents to unseen environments remains challenging. Recent methods explore pretraining to improve generalization, however, the use of generic image-caption datasets or existing small-scale VLN environments is suboptimal and results in limited improvements. In this work, we introduce BnB1, a large-scale and diverse in-domain VLN dataset. We first collect image-caption (IC) pairs from hundreds of thousands of listings from online rental marketplaces. Using IC pairs we next propose automatic strategies to generate millions of VLN path-instruction (PI) pairs. We further propose a shuffling loss that improves the learning of temporal order inside PI pairs. We use BnB to pretrain our Airbert2model that can be adapted to discriminative and generative settings and show that it outperforms state of the art for Room-to-Room (R2R) navigation and Remote Referring Expression (REVERIE) benchmarks. Moreover, our in-domain pretraining significantly increases performance on a challenging few-shot VLN evaluation, where we train the model only on VLN instructions from a few houses.

Image segmentation is often ambiguous at the level of individual image patches and requires contextual information to reach label consensus. In this paper we introduce Segmenter, a transformer model for semantic segmentation. In contrast to convolution-based methods, our approach allows to model global context already at the first layer and throughout the network. We build on the recent Vision Transformer (ViT) and extend it to semantic segmentation. To do so, we rely on the output embeddings corresponding to image patches and obtain class labels from these embed-dings with a point-wise linear decoder or a mask trans-former decoder. We leverage models pre-trained for image classification and show that we can fine-tune them on moderate sized datasets available for semantic segmentation. The linear decoder allows to obtain excellent results already, but the performance can be further improved by a mask transformer generating class masks. We conduct an extensive ablation study to show the impact of the different parameters, in particular the performance is better for large models and small patch sizes. Segmenter attains excellent results for semantic segmentation. It outperforms the state of the art on both ADE20K and Pascal Context datasets and is competitive on Cityscapes.

Recent methods for visual question answering rely on large-scale annotated datasets. Manual annotation of questions and answers for videos, however, is tedious, expensive and prevents scalability. In this work, we propose to avoid manual annotation and generate a large-scale training dataset for video question answering making use of automatic cross-modal supervision. We leverage a question generation transformer trained on text data and use it to generate question-answer pairs from transcribed video narrations. Given narrated videos, we then automatically generate the HowToVQA69M dataset with 69M video-question-answer triplets. To handle the open vocabulary of diverse answers in this dataset, we propose a training procedure based on a contrastive loss between a video-question multi-modal transformer and an answer transformer. We introduce the zero-shot VideoQA task and show excellent results, in particular for rare answers. Furthermore, we demonstrate our method to significantly outperform the state of the art on MSRVTT-QA, MSVD-QA, ActivityNet-QA and How2QA. Finally, for a detailed evaluation we introduce iVQA, a new VideoQA dataset with reduced language biases and high-quality redundant manual annotations.

Goal-conditioned reinforcement learning endows an agent with a large variety of skills, but it often struggles to solve tasks that require more temporally extended reasoning. In this work, we propose to incorporate imagined subgoals into policy learning to facilitate learning of complex tasks. Imagined subgoals are predicted by a separate high-level policy, which is trained simultaneously with the policy and its critic. This high-level policy predicts intermediate states halfway to the goal using the value function as a reachability metric. We don’t require the policy to reach these subgoals explicitly. Instead, we use them to define a prior policy, and incorporate this prior into a KL-constrained policy iteration scheme to speed up and regularize learning. Imagined subgoals are used during policy learning, but not during test time, where we only apply the learned policy. We evaluate our approach on complex robotic navigation and manipulation tasks and show that it outperforms existing methods by a large margin.

Following their success in natural language processing, transformers have recently shown much promise for computer vision. The self-attention operation underlying transformers yields global interactions between all tokens ,i.e. words or image patches, and enables flexible modelling of image data beyond the local interactions of convolutions. This flexibility, however, comes with a quadratic complexity in time and memory, hindering application to long sequences and high-resolution images. We propose a “transposed” version of self-attention that operates across feature channels rather than tokens, where the interactions are based on the cross-covariance matrix between keys and queries. The resulting cross-covariance attention (XCA) has linear complexity in the number of tokens, and allows efficient processing of high-resolution images.Our cross-covariance image transformer (XCiT) is built upon XCA. It combines the accuracy of conventional transformers with the scalability of convolutional architectures. We validate the effectiveness and generality of XCiT by reporting excellent results on multiple vision benchmarks, including image classification and self-supervised feature learning on ImageNet-1k, object detection and instance segmentation on COCO, and semantic segmentation on ADE20k.We will opensource our code and trained models to reproduce the reported results.

Vision-and-language navigation (VLN) aims to build autonomous visual agents that follow instructions and navigate in real scenes. To remember previously visited locations and actions taken, most approaches to VLN implement memory using recurrent states. Instead, we introduce a History Aware Multimodal Transformer (HAMT) to incorporate a long-horizon history into multimodal decision making. HAMT efficiently encodes all the past panoramic observations via a hierarchical vision transformer (ViT), which first encodes individual images with ViT, then models spatial relation between images in a panoramic observation and finally takes into account temporal relation between panoramas in the history. It, then, jointly combines text, history and current observation to predict the next action. We first train HAMT end-to-end using several proxy tasks including single step action prediction and spatial relation prediction, and then use reinforcement learning to further improve the navigation policy. HAMT achieves new state of the art on a broad range of VLN tasks, including VLN with fine-grained instructions (R2R, RxR), high-level instructions (R2R-Last, REVERIE), dialogs (CVDN) as well as long-horizon VLN (R4R, R2R-Back). We demonstrate HAMT to be particularly effective for navigation tasks with longer trajectories.

Reasoning about 3D scenes from their 2D image projections is one of the core problems in computer vision. Solutions to this inverse and ill-posed problem typically involve a search for models that best explain observed image data. Notably, images depend both on the properties of observed scenes and on the process of image formation. Hence, if optimization techniques should be used to explain images, it is crucial to design differentable functions for the projection of 3D scenes into images, also known as differentiable rendering. Previous approaches to differentiable rendering typically replace non-differentiable operations by smooth approximations, impacting the subsequent 3D estimation. In this paper, we take a more general approach and study differentiable renderers through the prism of randomized optimization and the related notion of perturbed optimizers. In particular, our work highlights the link between some well-known differentiable renderer formulations and randomly smoothed optimizers, and introduces differentiable perturbed renderers. We also propose a variance reduction mechanism to alleviate the computational burden inherent to perturbed optimizers and introduce an adaptive scheme to automatically adjust the smoothing parameters of the rendering process. We apply our method to 3D scene reconstruction and demonstrate its advantages on the tasks of 6D pose estimation and 3D mesh reconstruction. By providing informative gradients that can be used as a strong supervisory signal, we demonstrate the benefits of perturbed renderers to obtain more accurate solutions when compared to the state-of-the-art alternatives using smooth gradient approximations.

Abstract Although synthetic training data has been shown to be beneficial for tasks such as human pose estimation, its use for RGB human action recognition is relatively unexplored. Our goal in this work is to answer the question whether synthetic humans can improve the performance of human action recognition, with a particular focus on generalization to unseen viewpoints. We make use of the recent advances in monocular 3D human body reconstruction from real action sequences to automatically render synthetic training videos for the action labels. We make the following contributions: (1) we investigate the extent of variations and augmentations that are beneficial to improving performance at new viewpoints. We consider changes in body shape and clothing for individuals, as well as more action relevant augmentations such as non-uniform frame sampling, and interpolating between the motion of individuals performing the same action; (2) We introduce a new data generation methodology, SURREACT, that allows training of spatio-temporal CNNs for action classification; (3) We substantially improve the state-of-the-art action recognition performance on the NTU RGB+D and UESTC standard human action multi-view benchmarks; Finally, (4) we extend the augmentation approach to in-the-wild videos from a subset of the Kinetics dataset to investigate the case when only one-shot training data is available, and demonstrate improvements in this case as well.

Modeling hand-object manipulations is essential for understanding how humans interact with their environment. While of practical importance, estimating the pose of hands and objects during interactions is challenging due to the large mutual occlusions that occur during manipulation. Recent efforts have been directed towards fully-supervised methods that require large amounts of labeled training samples. Collecting 3D ground-truth data for hand-object interactions, however, is costly, tedious, and error-prone. To overcome this challenge we present a method to leverage photometric consistency across time when annotations are only available for a sparse subset of frames in a video. Our model is trained end-to-end on color images to jointly reconstruct hands and objects in 3D by inferring their poses. Given our estimated reconstructions, we differentiably render the optical flow between pairs of adjacent images and use it within the network to warp one frame to another. We then apply a self-supervised photometric loss that relies on the visual consistency between nearby images. We achieve state-of-the-art results on 3D hand-object reconstruction benchmarks and demonstrate that our approach allows us to improve the pose estimation accuracy by leveraging information from neighboring frames in low-data regimes.

Interactions between people are often governed by their relationships. On the flip side, social relationships are built upon several interactions. Two strangers are more likely to greet and introduce themselves while becoming friends over time. We are fascinated by this interplay between interactions and relationships, and believe that it is an important aspect of understanding social situations. In this work, we propose neural models to learn and jointly predict interactions, relationships, and the pair of characters that are involved. We note that interactions are informed by a mixture of visual and dialog cues, and present a multimodal architecture to extract meaningful information from them. Localizing the pair of interacting characters in video is a time-consuming process, instead, we train our model to learn from clip-level weak labels. We evaluate our models on the MovieGraphs dataset and show the impact of modalities, use of longer temporal context for predicting relationships, and achieve encouraging performance using weak labels as compared with ground-truth labels. Code is online.

Annotating videos is cumbersome, expensive and not scalable. Yet, many strong video models still rely on manually annotated data. With the recent introduction of the HowTo100M dataset, narrated videos now offer the possibility of learning video representations without manual supervision. In this work we propose a new learning approach, MIL-NCE, capable of addressing mis- alignments inherent in narrated videos. With this approach we are able to learn strong video representations from scratch, without the need for any manual annotation. We evaluate our representations on a wide range of four downstream tasks over eight datasets: action recognition (HMDB-51, UCF-101, Kinetics-700), text-to- video retrieval (YouCook2, MSR-VTT), action localization (YouTube-8M Segments, CrossTask) and action segmentation (COIN). Our method outperforms all published self-supervised approaches for these tasks as well as several fully supervised baselines.

Manipulation tasks such as preparing a meal or assembling furniture remain highly challenging for robotics and vision. Traditional task and motion planning (TAMP) methods can solve complex tasks but require full state observability and are not adapted to dynamic scene changes. Recent learning methods can operate directly on visual inputs but typically require many demonstrations and/or task-specific reward engineering. In this work we aim to overcome previous limitations and propose a reinforcement learning (RL) approach to task planning that learns to combine primitive skills. First, compared to previous learning methods, our approach requires neither intermediate rewards nor complete task demonstrations during training. Second, we demonstrate the versatility of our vision-based task planning in challenging settings with temporary occlusions and dynamic scene changes. Third, we propose an efficient training of basic skills from few synthetic demonstrations by exploring recent CNN architectures and data augmentation. Notably, while all of our policies are learned on visual inputs in simulated environments, we demonstrate the successful transfer and high success rates when applying such policies to manipulation tasks on a real UR5 robotic arm.

Manipulation and assembly tasks require non-trivial planning of actions depending on the environment and the final goal. Previous work in this domain often assembles particular instances of objects from known sets of primitives. In contrast, we aim to handle varying sets of primitives and to construct different objects of a shape category. Given a single object instance of a category, e.g. an arch, and a binary shape classifier, we learn a visual policy to assemble other instances of the same category. In particular, we propose a disassembly procedure and learn a state policy that discovers new object instances and their assembly plans in state space. We then render simulated states in the observation space and learn a heatmap representation to predict alternative actions from a given input image. To validate our approach, we first demonstrate its efficiency for building object categories in state space. We then show the success of our visual policies for building arches from different primitives. Moreover, we demonstrate (i) the reactive ability of our method to re-assemble objects using additional primitives and (ii) the robust performance of our policy for unseen primitives resembling building blocks used during training. Our visual assembly policies are trained with no real images and reach up to 95% success rate when evaluated on a real robot.

Estimating hand-object manipulations is essential for in- terpreting and imitating human actions. Previous work has made significant progress towards reconstruction of hand poses and object shapes in isolation. Yet, reconstructing hands and objects during manipulation is a more challeng- ing task due to significant occlusions of both the hand and object. While presenting challenges, manipulations may also simplify the problem since the physics of contact re- stricts the space of valid hand-object configurations. For example, during manipulation, the hand and object should be in contact but not interpenetrate. In this work, we regu- larize the joint reconstruction of hands and objects with ma- nipulation constraints. We present an end-to-end learnable model that exploits a novel contact loss that favors phys- ically plausible hand-object constellations. Our approach improves grasp quality metrics over baselines, using RGB images as input. To train and evaluate the model, we also propose a new large-scale synthetic dataset, ObMan, with hand-object manipulations. We demonstrate the transfer- ability of ObMan-trained models to real data.

Metric Learning for visual similarity has mostly adopted binary supervision indicating whether a pair of images are of the same class or not. Such a binary indicator covers only a limited subset of image relations, and is not sufficient to represent semantic similarity between images described by continuous and/or structured labels such as object poses, image captions, and scene graphs. Motivated by this, we present a novel method for deep metric learning using continuous labels. First, we propose a new triplet loss that allows distance ratios in the label space to be preserved in the learned metric space. The proposed loss thus enables our model to learn the degree of similarity rather than just the order. Furthermore, we design a triplet mining strategy adapted to metric learning with continuous labels. We address three different image retrieval tasks with continuous labels in terms of human poses, room layouts and image captions, and demonstrate the superior performance of our approach compared to previous methods.

In this paper, we introduce a method to automatically reconstruct the 3D motion of a person interacting with an object from a single RGB video. Our method estimates the 3D poses of the person and the object, contact positions, and forces and torques actuated by the human limbs. The main contributions of this work are three-fold. First, we introduce an approach to jointly estimate the motion and the actuation forces of the person on the manipulated object by modeling contacts and the dynamics of their interactions. This is cast as a large-scale trajectory optimization problem. Second, we develop a method to automatically recognize from the input video the position and timing of contacts between the person and the object or the ground, thereby significantly simplifying the complexity of the optimization. Third, we validate our approach on a recent MoCap dataset with ground truth contact forces and demonstrate its performance on a new dataset of Internet videos showing people manipulating a variety of tools in unconstrained environments.

In this paper we investigate learning visual models for the steps of ordinary tasks using weak supervision via instructional narrations and an ordered list of steps instead of strong supervision via temporal annotations. At the heart of our approach is the observation that weakly supervised learning may be easier if a model shares components while learning different steps: ``pour egg'' should be trained jointly with other tasks involving ``pour'' and ``egg''. We formalize this in a component model for recognizing steps and a weakly supervised learning framework that can learn this model under temporal constraints from narration and the list of steps. Past data does not permit systematic studying of sharing and so we also gather a new dataset aimed at assessing cross-task sharing. Our experiments demonstrate that sharing across tasks improves performance, especially when done at the component level and that our component model can parse previously unseen tasks by virtue of its compositionality.

Learning text-video embeddings usually requires a dataset of video clips with manually provided captions. However, such datasets are expensive and time consuming to create and therefore difficult to obtain on a large scale. In this work, we propose instead to learn such embeddings from video data with readily available natural language annotations in the form of automatically transcribed narrations. The contributions of this work are three-fold. First, we introduce HowTo100M: a large-scale dataset of 136 million video clips sourced from 1.22M narrated instructional web videos depicting humans performing and describing over 23k different visual tasks. Our data collection procedure is fast, scalable and does not require any additional manual annotation. Second, we demonstrate that a text-video embedding trained on this data leads to state-of-the-art results for text-to-video retrieval and action localization on instructional video datasets such as YouCook2 or CrossTask. Finally, we show that this embedding transfers well to other domains: fine-tuning on generic Youtube videos (MSR-VTT dataset) and movies (LSMDC dataset) outperforms models trained on these datasets alone. Our dataset, code and models are publicly available.

We seek to detect visual relations in images of the form of triplets t = (subject, predicate, object), such as “person riding dog”, where training examples of the individual entities are available but their combinations are unseen at training. This is an important set-up due to the combinatorial nature of visual relations : collecting sufficient training data for all possible triplets would be very hard. The contributions of this work are three-fold. First, we learn a representation of visual relations that combines (i) individual embeddings for subject, object and predicate together with (ii) a visual phrase embedding that represents the relation triplet. Second, we learn how to transfer visual phrase embeddings from existing training triplets to unseen test triplets using analogies between relations that involve similar objects. Third, we demonstrate the benefits of our approach on three challenging datasets : on HICO-DET, our model achieves significant improvement over a strong baseline for both frequent and unseen triplets, and we observe similar improvement for the retrieval of unseen triplets with out-of-vocabulary predicates on the COCO-a dataset as well as the challenging unusual triplets in the UnRel dataset.

With the rapid progress of face recognition it has more and more applications in everyday life. Although its backbone, very deep neural networks, also show improvement both in terms of accuracy and efficiency their computational cost and memory usage is still a limiting factor for deploying these models on a hardware with limited computational and power resources, such as mobile or embedded devices. Here arises the task of learning fast and compact deep neural networks which have a comparable accuracy to the complex model as requirement of real-life applications. Another issue is that sometimes face recognition system may run models of different complexity depending of the devices used for biometric template extraction (i.e. desktop with GPU or mobile phone), so the compatibility between the face descriptors is desirable. Our paper considers both this cases: we propose a new method for learning fast and compact face recognition model which has a similar performance to a much more complex model used for transferring its knowledge and we also show that both these models can be used for verification in a single face recognition system. To the best of our knowledge such evaluation of a compatibility between 2 different models for face recognition was never done before our work.

Vision and learning have made significant progress that could improve robotics policies for complex tasks and environments. Learning deep neural networks for image understanding, however, requires large amounts of domain-specific visual data. While collecting such data from real robots is possible, such an approach limits the scalability as learning policies typically requires thousands of trials. In this work we attempt to learn manipulation policies in simulated environments. Simulators enable scalability and provide access to the underlying world state during training. Policies learned in simulators, however, do not transfer well to real scenes given the domain gap between real and synthetic data. We follow recent work on domain randomization and augment synthetic images with sequences of random transformations. Our main contribution is to optimize the augmentation strategy for sim2real transfer and to enable domain-independent policy learning. We design an efficient search for depth image augmentations using object localization as a proxy task. Given the resulting sequence of random transformations, we use it to augment synthetic depth images during policy learning. Our augmentation strategy is policy-independent and enables policy learning with no real images. We demonstrate our approach to significantly improve accuracy on three manipulation tasks evaluated on a real robot.

Spatio-temporal action detection in videos is typically addressed in a fully-supervised setup with manual annotation of training videos required at every frame. Since such annotation is extremely tedious and prohibits scalability, there is a clear need to minimize the amount of manual supervision. In this work we propose a unifying framework that can handle and combine varying types of less demanding weak supervision. Our model is based on discriminative clustering and integrates different types of supervision as constraints on the optimization. We investigate applications of such a model to training setups with alternative supervisory signals ranging from video-level class labels over temporal points or sparse action bounding boxes to the full per-frame annotation of action bounding boxes. Experiments on the challenging UCF101-24 and DALY datasets demonstrate competitive performance of our method at a fraction of supervision used by previous methods. The flexibility of our model enables joint learning from data with different levels of annotation. Experimental results demonstrate a significant gain by adding a few fully supervised examples to otherwise weakly labeled videos.

Automatic assistants could guide a person or a robot in performing new tasks, such as changing a car tire or repotting a plant. Creating such assistants, however, is non-trivial and requires understanding of visual and verbal content of a video. Towards this goal, we here address the problem of automatically learning the main steps of a task from a set of narrated instruction videos. We develop a new unsupervised learning approach that takes advantage of the complementary nature of the input video and the associated narration. The method sequentially clusters textual and visual representations of a task, where the two clustering problems are linked by joint constraints to obtain a single coherent sequence of steps in both modalities. To evaluate our method, we collect and annotate a new challenging dataset of real-world instruction videos from the Internet. The dataset contains videos for five different tasks with complex interactions between people and objects, captured in a variety of indoor and outdoor settings. We experimentally demonstrate that the proposed method can automatically discover, learn and localize the main steps of a task in input videos.

Typical human actions last several seconds and exhibit characteristic spatio-temporal structure. Recent methods attempt to capture this structure and learn action representations with convolutional neural networks. Such representations, however, are typically learned at the level of a few video frames failing to model actions at their full temporal extent. In this work we learn video representations using neural networks with long-term temporal convolutions (LTC). We demonstrate that LTC-CNN models with increased temporal extents improve the accuracy of action recognition. We also study the impact of different low-level representations, such as raw values of video pixels and optical flow vector fields and demonstrate the importance of high-quality optical flow estimation for learning accurate action models. We report state-of-the-art results on two challenging benchmarks for human action recognition UCF101 (92.7%) and HMDB51 (67.2%).

Estimating human pose, shape, and motion from images and videos are fundamental challenges with many applications. Recent advances in 2D human pose estimation use large amounts of manually-labeled training data for learning convolutional neural networks (CNNs). Such data is time consuming to acquire and difficult to extend. Moreover, manual labeling of 3D pose, depth and motion is impractical. In this work we present SURREAL (Synthetic hUmans foR REAL tasks): a new large-scale dataset with synthetically-generated but realistic images of people rendered from 3D sequences of human motion capture data. We generate more than 6 million frames together with ground truth pose, depth maps, and segmentation masks. We show that CNNs trained on our synthetic dataset allow for accurate human depth estimation and human part segmentation in real RGB images. Our results and the new dataset open up new possibilities for advancing person analysis using cheap and large-scale synthetic data.

Many human activities involve object manipulations aiming to modify the object state. Examples of common state changes include full/empty bottle, open/closed door, and attached/detached car wheel. In this work, we seek to automatically discover the states of objects and the associated manipulation actions. Given a set of videos for a particular task, we propose a joint model that learns to identify object states and to localize state-modifying actions. Our model is formulated as a discriminative clustering cost with constraints. We assume a consistent temporal order for the changes in object states and manipulation actions, and introduce new optimization techniques to learn model parameters without additional supervision. We demonstrate successful discovery of seven manipulation actions and corresponding object states on a new dataset of videos depicting real-life object manipulations. We show that our joint formulation results in an improvement of object state discovery by action recognition and vice versa.

Discriminative clustering has been successfully applied to a number of weakly supervised learning tasks. Such applications include person and action recognition, text-to-video alignment, object co-segmentation and co-localization in videos and images. One drawback of discriminative clustering, however, is its limited scalability. We address this issue and propose an online optimization algorithm based on the Block-Coordinate Frank-Wolfe algorithm. We apply the proposed method to the problem of weakly supervised learning of actions and actors from movies together with corresponding movie scripts. The scaling up of the learning problem to 66 feature-length movies enables us to significantly improve weakly supervised action recognition.

This paper introduces a novel approach for modeling visual relations between pairs of objects. We call relation a triplet of the form (subject; predicate; object) where the predicate is typically a preposition (eg. 'under', 'in front of') or a verb ('hold', 'ride') that links a pair of objects (subject; object). Learning such relations is challenging as the objects have different spatial configurations and appearances depending on the relation in which they occur. Another major challenge comes from the difficulty to get annotations, especially at box-level, for all possible triplets, which makes both learning and evaluation difficult. The contributions of this paper are threefold. First, we design strong yet flexible visual features that encode the appearance and spatial configuration for pairs of objects. Second, we propose a weakly-supervised discriminative clustering model to learn relations from image-level labels only. Third we introduce a new challenging dataset of unusual relations (UnRel) together with an exhaustive annotation, that enables accurate evaluation of visual relation retrieval. We show experimentally that our model results in state-of-the-art results on the visual relationship dataset [32] significantly improving performance on previously unseen relations (zero-shot learning), and confirm this observation on our newly introduced UnRel dataset.

We address the problem of automatically learning the main steps to complete a certain task, such as changing a car tire, from a set of narrated instruction videos. The contributions of this paper are three-fold. First, we develop a new unsupervised learning approach that takes advantage of the complementary nature of the input video and the associated narration. The method solves two clustering problems, one in text and one in video, applied one after each other and linked by joint constraints to obtain a single coherent sequence of steps in both modalities. Second, we collect and annotate a new challenging dataset of real-world instruction videos from the Internet. The dataset contains about 800,000 frames for five different tasks1that include complex interactions between people and objects, and are captured in a variety of indoor and outdoor settings. Third, we experimentally demonstrate that the proposed method can automatically discover, in an unsupervised manner, the main steps to achieve the task and locate the steps in the input videos.

We address the problem of learning a pose-aware, compact embedding that projects images with similar human poses to be placed close-by in the embedding space. The embedding function is built on a deep convolutional network, and trained with triplet-based rank constraints on real image data. This architecture allows us to learn a robust representation that captures differences in human poses by effectively factoring out variations in clothing, background, and imaging conditions in the wild. For a variety of pose-related tasks, the proposed pose embedding provides a cost-efficient and natural alternative to explicit pose estimation, circumventing challenges of localizing body joints. We demonstrate the efficacy of the embedding on pose-based image retrieval and action recognition problems.

We address the problem of segmenting multiple object instances in complex videos. Our method does not require manual pixel-level annotation for training, and relies instead on readily-available object detectors or visual object tracking only. Given object bounding boxes at input, we cast video segmentation as a weakly-supervised learning problem. Our proposed objective combines (a) a discriminative clustering term for background segmentation, (b) a spectral clustering one for grouping pixels of same object instances, and (c) linear constraints enabling instance-level segmentation. We propose a convex relaxation of this problem and solve it efficiently using the Frank-Wolfe algorithm. We report results and compare our method to several baselines on a new video dataset for multi-instance person segmentation.

Large-scale annotated datasets allow AI systems to learn from and build upon the knowledge of the crowd. Many crowdsourcing techniques have been developed for collecting image annotations. These techniques often implicitly rely on the fact that a new input image takes a negligible amount of time to perceive. In contrast, we investigate and determine the most cost-effective way of obtaining high-quality multi-label annotations for temporal data such as videos. Watching even a short 30-second video clip requires a significant time investment from a crowd worker; thus, requesting multiple annotations following a single viewing is an important cost-saving strategy. But how many questions should we ask per video? We conclude that the optimal strategy is to ask as many questions as possible in a HIT (up to 52 binary questions after watching a 30-second video clip in our experiments).We demonstrate that while workers may not correctly answer all questions, the cost-benefit analysis nevertheless favors consensus from multiple such cheap-yet-imperfect iterations over more complex alternatives. When compared with a one-question-per-video baseline, our method is able to achieve a 10% improvement in recall (76.7% ours versus 66.7% baseline) at comparable precision (83.8% ours versus 83.0% baseline) in about half the annotation time (3.8 minutes ours compared to 7.1 minutes baseline). We demonstrate the effectiveness of our method by collecting multi-label annotations of 157 human activities on 1,815 videos.

Multi-object tracking has been recently approached with the min-cost network flow optimization techniques. Such methods simultaneously resolve multiple object tracks in a video and enable modeling of dependencies among tracks. Min-cost network flow methods also fit well within the “tracking-by-detection” paradigm where object trajectories are obtained by connecting per-frame outputs of an object detector. Object detectors, however, often fail due to occlusions and clutter in the video. To cope with such situations, we propose to add pairwise costs to the min-cost network flow framework. While integer solutions to such a problem become NP-hard, we design a convex relaxation solution with an efficient rounding heuristic which empirically gives certificates of small suboptimality. We evaluate two particular types of pairwise costs and demonstrate improvements over recent tracking methods in real-world video sequences.

Successful methods for visual object recognition typically rely on training datasets containing lots of richly annotated images. Detailed image annotation, e.g. by object bounding boxes, however, is both expensive and often subjective. We describe a weakly supervised convolutional neural network (CNN) for object classification that relies only on image-level labels, yet can learn from cluttered scenes containing multiple objects. We quantify its object classification and object location prediction performance on the Pascal VOC 2012 (20 object classes) and the much larger Microsoft COCO (80 object classes) datasets. We find that the network (i) outputs accurate image-level labels, (ii) predicts approximate locations (but not extents) of objects, and (iii) performs comparably to its fully-supervised counterparts using object bounding box annotation for training.

Suppose that we are given a set of videos, along with natural language descriptions in the form of multiple sentences (e.g., manual annotations, movie scripts, sport summaries etc.), and that these sentences appear in the same temporal order as their visual counterparts. We propose in this paper a method for aligning the two modalities, i.e., automatically providing a time (frame) stamp for every sentence. Given vectorial features for both video and text, this can be cast as a temporal assignment problem, with an implicit linear mapping between the two feature modalities. We formulate this problem as an integer quadratic program, and solve its continuous convex relaxation using an efficient conditional gradient algorithm. Several rounding procedures are proposed to construct the final integer solution. After demonstrating significant improvements over the state of the art on the related task of aligning video with symbolic labels [7], we evaluate our method on a challenging dataset of videos with associated textual descriptions [37], and explore bag-of-words and continuous representations for text.

This work targets human action recognition in video. While recent methods typically represent actions by statistics of local video features, here we argue for the importance of a representation derived from human pose. To this end we propose a new Pose-based Convolutional Neural Network descriptor (P-CNN) for action recognition. The descriptor aggregates motion and appearance information along tracks of human body parts. We investigate different schemes of temporal aggregation and experiment with P-CNN features obtained both for automatically estimated and manually annotated human poses. We evaluate our method on the recent and challenging JHMDB and MPII Cooking datasets. For both datasets our method shows consistent improvement over the state of the art.

This paper addresses the problem of automatically localizing dominant objects as spatio-temporal tubes in a noisy collection of videos with minimal or even no supervision. We formulate the problem as a combination of two complementary processes: discovery and tracking. The first one establishes correspondences between prominent regions across videos, and the second one associates similar object regions within the same video. Interestingly, our algorithm also discovers the implicit topology of frames associated with instances of the same object class across different videos, a role normally left to supervisory information in the form of class labels in conventional image and video understanding methods. Indeed, as demonstrated by our experiments, our method can handle video collections featuring multiple object classes, and substantially outperforms the state of the art in colocalization, even though it tackles a broader problem with much less supervision.

Person detection is a key problem for many computer vision tasks. While face detection has reached maturity, detecting people under full variation of camera view-points, human poses, lighting conditions and occlusions is still a difficult challenge. In this work we focus on detecting human heads in natural scenes. Starting from the recent R-CNN object detector, we extend it in two ways. First, we leverage person-scene relations and propose a global CNN model trained to predict positions and scales of heads directly from the full image. Second, we explicitly model pairwise relations among the objects via energy-based model where the potentials are computed with a CNN framework. Our full combined model complements R-CNN with contextual cues derived from the scene. To train and test our model, we introduce a large dataset with 369,846 human heads annotated in 224,740 movie frames. We evaluate our method and demonstrate improvements of person head detection compared to several recent baselines on three datasets. We also show improvements of the detection speed provided by our model.

We describe a method to obtain a pixel-wise segmentation and pose estimation of multiple people in stereoscopic videos. This task involves challenges such as dealing with unconstrained stereoscopic video, non-stationary cameras, and complex indoor and outdoor dynamic scenes with multiple people. We cast the problem as a discrete labelling task involving multiple person labels, devise a suitable cost function, and optimize it efficiently. The contributions of our work are two-fold: First, we develop a segmentation model incorporating person detections and learnt articulated pose segmentation masks, as well as colour, motion, and stereo disparity cues. The model also explicitly represents depth ordering and occlusion. Second, we introduce a stereoscopic dataset with frames extracted from feature-length movies "StreetDance 3D" and "Pina". The dataset contains 587 annotated human poses, 1,158 bounding box annotations and 686 pixel-wise segmentations of people. The dataset is composed of indoor and outdoor scenes depicting multiple people with frequent occlusions. We demonstrate results on our new challenging dataset, as well as on the H2view dataset from (Sheasby et al. ACCV 2012).

Local video features provide state-of-the-art performance for action recognition. While the accuracy of action recognition has been continuously improved over the recent years, the low speed of feature extraction and subsequent recognition prevents current methods from scaling up to real-size problems. We address this issue and first develop highly efficient video features using motion information in video compression. We next explore feature encoding by Fisher vectors and demonstrate accurate action recognition using fast linear classifiers. Our method improves the speed of video feature extraction, feature encoding and action classification by two orders of magnitude at the cost of minor reduction in recognition accuracy. We validate our approach and compare it to the state of the art on four recent action recognition datasets.

Convolutional neural networks (CNN) have recently shown outstanding image classification performance in the large- scale visual recognition challenge (ILSVRC2012). The success of CNNs is attributed to their ability to learn rich mid-level image representations as opposed to hand-designed low-level features used in other image classification methods. Learning CNNs, however, amounts to estimating millions of parameters and requires a very large number of annotated image samples. This property currently prevents application of CNNs to problems with limited training data. In this work we show how image representations learned with CNNs on large-scale annotated datasets can be efficiently transferred to other visual recognition tasks with limited amount of training data. We design a method to reuse layers trained on the ImageNet dataset to compute mid-level image representation for images in the PASCAL VOC dataset. We show that despite differences in image statistics and tasks in the two datasets, the transferred representation leads to significantly improved results for object and action classification, outperforming the current state of the art on Pascal VOC 2007 and 2012 datasets. We also show promising results for object and action localization.

We seek to obtain a pixel-wise segmentation and pose estimation of multiple people in a stereoscopic video. This involves challenges such as dealing with unconstrained stereoscopic video, non-stationary cameras, and complex indoor and outdoor dynamic scenes. The contributions of our work are two-fold: First, we develop a segmentation model incorporating person detection, pose estimation, as well as colour, motion, and disparity cues. Our new model explicitly represents depth ordering and occlusion. Second, we introduce a stereoscopic dataset with frames extracted from feature-length movies "Street Dance 3D" and "Pina". The dataset contains 2727 realistic stereo pairs and includes annotation of human poses, person bounding boxes, and pixel-wise segmentations for hundreds of people. The dataset is composed of indoor and outdoor scenes depicting multiple people with frequent occlusions. We demonstrate results on our new challenging dataset, as well as on the H2view dataset from (Sheasby et al. ACCV 2012).

We address the problem of learning a joint model of actors and actions in movies using weak supervision provided by scripts. Specifically, we extract actor/action pairs from the script and use them as constraints in a discriminative clustering framework. The corresponding optimization problem is formulated as a quadratic program under linear constraints. People in video are represented by automatically extracted and tracked faces together with corresponding motion features. First, we apply the proposed framework to the task of learning names of characters in the movie and demonstrate significant improvements over previous methods used for this task. Second, we explore the joint actor/action constraint and show its advantage for weakly supervised action learning. We validate our method in the challenging setting of localizing and recognizing characters and their actions in feature length movies Casablanca and American Beauty.

This paper addresses the problem of human action recognition in realistic videos. We follow the recently successful local approaches and represent videos by means of local motion descriptors. To overcome the huge variability of human actions in motion and appearance, we propose a supervised approach to learn local motion descriptors - actlets - from a large pool of annotated video data. The main motivation behind our method is to construct action-characteristic representations of body joints undergoing specific motion patterns while learning invariance with respect to changes in camera views, lighting, human clothing, and other factors. We avoid the prohibitive cost of manual supervision and show how to learn actlets automatically from synthetic videos of avatars driven by the motion-capture data. We evaluate our method and show its improvement as well as its complementarity to existing techniques on the challenging UCF-Sports and YouTube-Actions datasets.

Video provides not only rich visual cues such as motion and appearance, but also much less explored long-range temporal interactions among objects. We aim to capture such interactions and to construct a powerful intermediate-level video representation for subsequent recognition. Motivated by this goal, we seek to obtain spatio-temporal oversegmentation of a video into regions that respect object boundaries and, at the same time, associate object pixels over many video frames. The contributions of this paper are two-fold. First, we develop an efficient spatiotemporal video segmentation algorithm, which naturally incorporates long-range motion cues from the past and future frames in the form of clusters of point tracks with coherent motion. Second, we devise a new track clustering cost function that includes occlusion reasoning, in the form of depth ordering constraints, as well as motion similarity along the tracks. We evaluate the proposed approach on a challenging set of video sequences of office scenes from feature length movies.

We address the problem of person detection and tracking in crowded video scenes. While the detection of individual objects has been improved significantly over the recent years, crowd scenes remain particularly challenging for the detection and tracking tasks due to heavy occlusions, high person densities and significant variation in people's appearance. To address these challenges, we propose to leverage information on the global structure of the scene and to resolve all detections jointly. In particular, we explore constraints imposed by the crowd density and formulate person detection as the optimization of a joint energy function combining crowd density estimation and the localization of individual people. We demonstrate how the optimization of such an energy function significantly improves person detection and tracking in crowds. We validate our approach on a challenging video dataset of crowded scenes.

In this work we present a new crowd analysis algorithm powered by behavior priors that are learned on a large database of crowd videos gathered from the Internet. The algorithm works by first learning a set of crowd behavior priors off-line. During testing, crowd patches are matched to the database and behavior priors are transferred. We adhere to the insight that despite the fact that the entire space of possible crowd behaviors is infinite, the space of distinguishable crowd motion patterns may not be all that large. For many individuals in a crowd, we are able to find analogous crowd patches in our database which contain similar patterns of behavior that can effectively act as priors to constrain the difficult task of tracking an individual in a crowd. Our algorithm is data-driven and, unlike some crowd characterization methods, does not require us to have seen the test video beforehand. It performs like state-of-the-art methods for tracking people having common crowd behaviors and outperforms the methods when the tracked individual behaves in an unusual way.

Human analysis in images and video is a hard problem due to the large variation in human pose, clothing, camera view-points, lighting and other factors. While the explicit modeling of this variability is difficult, the huge amount of available person images motivates for the implicit, data-driven approach to human analysis. In this work we aim to explore this approach using the large amount of images spanning a subspace of human appearance. We model this subspace by connecting images into a graph and propagating information through such a graph using a discriminatively-trained graphical model. We particularly address the problems of human pose estimation and action recognition and demonstrate how image graphs help solving these problems jointly. We report results on still images with human actions from the KTH dataset.

We investigate a discriminatively trained model of person-object interactions for recognizing common human actions in still images. We build on the locally order-less spatial pyramid bag-of-features model, which was shown to perform extremely well on a range of object, scene and human action recognition tasks. We introduce three principal contributions. First, we replace the standard quantized local HOG/SIFT features with stronger discriminatively trained body part and object detectors. Second, we introduce new person-object interaction features based on spatial co-occurrences of individual body parts and objects. Third, we address the combinatorial problem of a large number of possible interaction pairs and propose a discriminative selection procedure using a linear support vector machine (SVM) with a sparsity inducing regularizer. Learning of action-specific body part and object interactions bypasses the difficult problem of estimating the complete human body pose configuration. Benefits of the proposed model are shown on human action recognition in consumer photographs, outperforming the strong bag-of-features baseline.

This paper addresses recognition of human actions under view changes. We explore self-similarities of action sequences over time and observe the striking stability of such measures across views. Building upon this key observation, we develop an action descriptor that captures the structure of temporal similarities and dissimilarities within an action sequence. Despite this temporal self-similarity descriptor not being strictly view-invariant, we provide intuition and experimental validation demonstrating its high stability under view changes. Self-similarity descriptors are also shown to be stable under performance variations within a class of actions when individual speed fluctuations are ignored. If required, such fluctuations between two different instances of the same action class can be explicitly recovered with dynamic time warping, as will be demonstrated, to achieve cross-view action synchronization. More central to the current work, temporal ordering of local self-similarity descriptors can simply be ignored within a bag-of-features type of approach. Sufficient action discrimination is still retained in this way to build a view-independent action recognition system. Interestingly, self-similarities computed from different image features possess similar properties and can be used in a complementary fashion. Our method is simple and requires neither structure recovery nor multiview correspondence estimation. Instead, it relies on weak geometric properties and combines them with machine learning for efficient cross-view action recognition. The method is validated on three public data sets. It has similar or superior performance compared to related methods and it performs well even in extreme conditions, such as when recognizing actions from top views while using side views only for training.

International audience

Local space-time features have recently shown promising results within Bag-of-Features (BoF) approach to action recognition in video. Pure local features and descriptors, however, provide only limited discriminative power implying ambiguity among features and sub-optimal classification performance. In this work, we propose to disambiguate local space-time features and to improve action recognition by integrating additional nonlocal cues with BoF representation. For this purpose, we decompose video into region classes and augment local features with corresponding region-class labels. In particular, we investigate unsupervised and supervised video segmentation using (i) motion-based foreground segmentation, (ii) person detection, (iii) static action detection and (iv) object detection. While such segmentation methods might be imperfect, they provide complementary region-level information to local features. We demonstrate how this information can be integrated with BoF representations in a kernel-combination framework. We evaluate our method on the recent and challenging Hollywood-2 action dataset and demonstrate significant improvements.

This paper deals with the temporal synchronization of image sequences. Two instances of this problem are considered: (a) synchronization of human actions and (b) synchronization of dynamic scenes with view changes. To address both tasks and to reliably handle large view variations, we use self-similarity matrices which remain stable across views. We propose time-adaptive descriptors that capture the structure of these matrices while being invariant to the impact of time warps between views. Synchronizing two sequences is then performed by aligning their temporal descriptors using the Dynamic Time Warping algorithm. We present quantitative comparison results between time-fixed and time-adaptive descriptors for image sequences with different frame rates. We also illustrate the performance of the approach on several challenging videos with large view variations, drastic independent camera motions and within-class variability of human actions.

International audience

This paper exploits the context of natural dynamic scenes for human action recognition in video. Human actions are frequently constrained by the purpose and the physical properties of scenes and demonstrate high correlation with particular scene classes. For example, eating often happens in a kitchen while running is more common outdoors. The contribution of this paper is three-fold: (a) we automatically discover relevant scene classes and their correlation with human actions, (b) we show how to learn selected scene classes from video without manual supervision and (c) we develop a joint framework for action and scene recognition and demonstrate improved recognition of both in natural video. We use movie scripts as a means of automatic supervision for training. For selected action classes we identify correlated scene classes in text and then retrieve video samples of actions and scenes for training using script-to-video alignment. Our visual models for scenes and actions are formulated within the bag-of-features framework and are combined in a joint scene-action SVM-based classifier. We report experimental results and validate the method on a new large dataset with twelve action classes and ten scene classes acquired from 69 movies.

This paper addresses the problem of automatic temporal annotation of realistic human actions in video using minimal manual supervision. To this end we consider two associated problems: (a) weakly-supervised learning of action models from readily available annotations, and (b) temporal localization of human actions in test videos. To avoid the prohibitive cost of manual annotation for training, we use movie scripts as a means of weak supervision. Scripts, however, provide only implicit, noisy, and imprecise information about the type and location of actions in video. We address this problem with a kernel-based discriminative clustering algorithm that locates actions in the weakly-labeled training data. Using the obtained action samples, we train temporal action detectors and apply them to locate actions in the raw video data. Our experiments demonstrate that the proposed method for weakly-supervised learning of action models leads to significant improvement in action detection. We present detection results for three action classes in four feature length movies with challenging and realistic video data.

The aim of this paper is to address recognition of natural human actions in diverse and realistic video settings. This challenging but important subject has mostly been ignored in the past due to several problems one of which is the lack of realistic and annotated video datasets. Our first contribution is to address this limitation and to investigate the use of movie scripts for automatic annotation of human actions in videos. We evaluate alternative methods for action retrieval from scripts and show benefits of a text-based classifier. Using the retrieved action samples for visual learning, we next turn to the problem of action classification in video. We present a new method for video classification that builds upon and extends several recent ideas including local space-time features, space-time pyramids and multi-channel non-linear SVMs. The method is shown to improve state-of-the-art results on the standard KTH action dataset by achieving 91.8% accuracy. Given the inherent problem of noisy labels in automatic annotation, we particularly investigate and show high tolerance of our method to annotation errors in the training set. We finally apply the method to learning and classifying challenging action classes in movies and show promising results.

We address recognition and localization of human actions in realistic scenarios. In contrast to the previous work studying human actions in controlled settings, here we train and test algorithms on real movies with substantial variation of actions in terms of subject appearance, motion, surrounding scenes, viewing angles and spatio-temporal extents. We introduce a new annotated human action dataset and use it to evaluate several existing methods. We in particular focus on boosted space-time window classifiers and introduce "keyframe priming" that combines discriminative models of human motion and shape within an action. Keyframe priming is shown to significantly improve the performance of action detection. We present detection results for the action class "drinking" evaluated on two episodes of the movie "Coffee and Cigarettes".

We present a method for object detection that combines AdaBoost learning with local histogram features. On the side of learning we improve the performance by designing a weak learner for multi-valued features based on Weighted Fisher Linear Discriminant. Evaluation on the recent benchmark for object detection confirms the superior performance of our method compared to the state-of-the-art. In particular, using a single set of parameters our approach outperforms all methods reported in [5] for 7 out of 8 detection tasks and four object classes. 1

A method for detecting and segmenting periodic motion is presented. We exploit periodicity as a cue and detect periodic motion in complex scenes where common methods for motion segmentation are likely to fail. We note that periodic motion detection can be seen as an approximate case of sequence alignment where an image sequence is matched to itself over one or more periods of time. To use this observation, we first consider alignment of two video sequences obtained by independently moving cameras. Under assumption of constant translation, the fundamental matrices and the homographies are shown to be time-linear matrix functions. These dynamic quantities can be estimated by matching corresponding space-time points with similar local motion and shape. For periodic motion, we match corresponding points across periods and develop a RANSAC procedure to simultaneously estimate the period and the dynamic geometric transformations between periodic views. Using this method, we demonstrate detection and segmentation of human periodic motion in complex scenes with nonrigid backgrounds, moving camera and motion parallax.

Local image features or interest points provide compact and abstract representations of patterns in an image. We propose to extend the notion of spatial interest points into the spatio-temporal domain and show how the resulting features often reflect interesting events that can be used for a compact representation of video data as well as for its interpretation. To detect spatio-temporal events, we build on the idea of the Harris and Forstner interest point operators and detect local structures in space-time where the image values have significant local variations in both space and time. We then estimate the spatio-temporal extents of the detected events and compute their scale-invariant spatio-temporal descriptors. Using such descriptors, we classify events and construct video representation in terms of labeled space-time points. For the problem of human motion analysis, we illustrate how the proposed method allows for detection of walking people in scenes with occlusions and dynamic backgrounds.