Real-world scenes often feature multiple humans interacting with multiple objects in ways that are causal, goal-oriented, or cooperative. Yet existing 3D human-object interaction (HOI) benchmarks consider only a fraction of these complex interactions. To close this gap, we present MMHOI -- a large-scale, Multi-human Multi-object Interaction dataset consisting of images from 12 everyday scenarios. MMHOI offers complete 3D shape and pose annotations for every person and object, along with labels for 78 action categories and 14 interaction-specific body parts, providing a comprehensive testbed for next-generation HOI research. Building on MMHOI, we present MMHOI-Net, an end-to-end transformer-based neural network for jointly estimating human-object 3D geometries, their interactions, and associated actions. A key innovation in our framework is a structured dual-patch representation for modeling objects and their interactions, combined with action recognition to enhance the interaction prediction. Experiments on MMHOI and the recently proposed CORE4D datasets demonstrate that our approach achieves state-of-the-art performance in multi-HOI modeling, excelling in both accuracy and reconstruction quality. The MMHOI dataset is publicly available at https://zenodo.org/records/17711786.

Assembling furniture amounts to solving the discrete-continuous optimization task of selecting the furniture parts to assemble and estimating their connecting poses in a physically realistic manner. The problem is hampered by its combinatorially large yet sparse solution space thus making learning to assemble a challenging task for current machine learning models. In this paper, we attempt to solve this task by leveraging the assembly instructions provided in diagrammatic manuals that typically accompany the furniture parts. Our key insight is to use the cues in these diagrams to split the problem into discrete and continuous phases. Specifically, we present Manual-PA, a transformer-based instruction Manual-guided 3D Part Assembly framework that learns to semantically align 3D parts with their illustrations in the manuals using a contrastive learning backbone towards predicting the assembly order and infers the 6D pose of each part via relating it to the final furniture depicted in the manual. To validate the efficacy of our method, we conduct experiments on the benchmark PartNet dataset. Our results show that using the diagrams and the order of the parts lead to significant improvements in assembly performance against the state of the art. Further, Manual-PA demonstrates strong generalization to real-world IKEA furniture assembly on the IKEA-Manual dataset.

Accurately localizing 3D sound sources and estimating their semantic labels - where the sources may not be visible, but are assumed to lie on the physical surface of objects in the scene - have many real applications, including detecting gas leak and machinery malfunction. The audio-visual weak-correlation in such setting poses new challenges in deriving innovative methods to answer if or how we can use cross-modal information to solve the task. Towards this end, we propose to use an acoustic-camera rig consisting of a pinhole RGB-D camera and a coplanar four-channel microphone array (Mic-Array). By using this rig to record audio-visual signals from multiviews, we can use the cross-modal cues to estimate the sound sources 3D locations. Specifically, our framework SoundLoc3D treats the task as a set prediction problem, each element in the set corresponds to a potential sound source. Given the audio-visual weak-correlation, the set representation is initially learned from a single view microphone array signal, and then refined by actively incorporating physical surface cues revealed from multiview RGB-D images. We demonstrate the efficiency and superiority of SoundLoc3D on large-scale simulated dataset, and further show its robustness to RGB-D measurement inaccuracy and ambient noise interference.

Audio-visual navigation of an agent towards locating an audio goal is a challenging task especially when the audio is sporadic or the environment is noisy. In this paper, we present CAVEN, a Conversation-based Audio-Visual Embodied Navigation framework in which the agent may interact with a human/oracle for solving the task of navigating to an audio goal. Specifically, CAVEN is modeled as a budget-aware partially observable semi-Markov decision process that implicitly learns the uncertainty in the audio-based navigation policy to decide when and how the agent may interact with the oracle. Our CAVEN agent can engage in fully-bidirectional natural language conversations by producing relevant questions and interpret free-form, potentially noisy responses from the oracle based on the audio-visual context. To enable such a capability, CAVEN is equipped with: i) a trajectory forecasting network that is grounded in audio-visual cues to produce a potential trajectory to the estimated goal, and (ii) a natural language based question generation and reasoning network to pose an interactive question to the oracle or interpret the oracle's response to produce navigation instructions. To train the interactive modules, we present a large scale dataset: AVN-Instruct, based on the Landmark-RxR dataset. To substantiate the usefulness of conversations, we present experiments on the benchmark audio-goal task using the SoundSpaces simulator under various noisy settings. Our results reveal that our fully-conversational approach leads to nearly an order-of-magnitude improvement in success rate, especially in localizing new sound sources and against methods that use only uni-directional interaction.

Text-conditioned image-to-video generation (TI2V) aims to synthesize a realistic video starting from a given image (e.g., a woman's photo) and a text description (e.g., “a woman is drinking water.”). Existing TI2V frameworks often require costly training on video-text datasets and spe-cific model designs for text and image conditioning. In this paper, we propose TI2V-Zero, a zero-shot, tuning-free method that empowers a pretrained text-to-video (T2V) diffusion model to be conditioned on a provided image, enabling TI2V generation without any optimization, fine-tuning, or introducing external modules. Our approach leverages a pretrained T2V diffusion foundation model as the generative prior. To guide video generation with the additional image input, we propose a “repeat-and-slide” strategy that modulates the reverse denoising process, al-lowing the frozen diffusion model to synthesize a video frame-by-frame starting from the provided image. To ensure temporal continuity, we employ a DDPM inversion strategy to initialize Gaussian noise for each newly synthesized frame and a resampling technique to help preserve visual details. We conduct comprehensive experiments on both domain-specific and open-domain datasets, where TI2V-Zero consistently outperforms a recent open-domain TI2V model. Furthermore, we show that TI2V-Zero can seam-lessly extend to other tasks such as video infilling and pre-diction when provided with more images. Its autoregressive design also supports long video generation.

We leverage Large Language Models (LLM) for zero-shot Semantic Audio Visual Navigation (SAVN). Existing methods utilize extensive training demonstrations for rein-forcement learning, yet achieve relatively low success rates and lack generalizability. The intermittent nature of au-ditory signals further poses additional obstacles to infer-ring the goal information. To address this challenge, we present the Reflective and Imaginative Language Agent (RILA). By employing multi-modal models to process sen-sory data, we instruct an LLM-based planner to actively ex-plore the environment. During the exploration, our agent adaptively evaluates and dismisses inaccurate perceptual descriptions. Additionally, we introduce an auxiliary LLM-based assistant to enhance global environmental compre-hension by mapping room layouts and providing strategic insights. Through comprehensive experiments and analy-sis, we show that our method outperforms relevant base-lines without training demonstrations from the environment and complementary semantic information.

Indoor monitoring systems are in high demand to protect vulnerable people, especially when they are alone at home, in nursing homes, hospitals, etc. Although surveillance systems in public spaces use cameras and microphones to find incidents, indoor monitoring in personal spaces needs to protect privacy. Such systems thus need to understand scenes without relying on direct sensing information, e.g., from audio-visual sensors, instead using indirect sensing information that is difficult to interpret by humans and may be insufficient to understand ongoing events precisely. To mitigate this drawback, this paper proposes a new indoor monitoring approach that attempts to realize scene understanding using only indirect sensors by transferring the learned inductive bias of a multimodal fusion model trained using direct and indirect sensing information to a model that uses only indirect information during inference. We collected direct (audio-visual) and indirect (infrared and Wi-Fi) sensing information of indoor human actions in daily life and manually annotated event captions. We build models that can generate event captions from various combinations of indirect and direct sensor data, and show that our transfer learning approach leads to significant improvements in caption quality when only indirect information is used at inference time.

The primary objective of room acoustics is to model the intricate sound propagation dynamics from any source to receiver position within enclosed 3D spaces. These dynamics are encapsulated in the form of a 1D room impulse response (RIR). Precisely measuring RIR is difficult due to the complexity of sound propagation encompassing reflection, diffraction, and absorption. In this work, we propose to learn a continuous neural room acoustics field that implicitly encodes all essential sound propagation primitives for each enclosed 3D space, so that we can infer the RIR corresponding to arbitrary source-receiver positions unseen in the training dataset. Our framework, dubbed DeepNeRAP, is trained in a self-supervised manner without requiring direct access to RIR ground truth that is often needed in prior methods. The key idea is to design two cooperative acoustic agents to actively probe a 3D space, one emitting and the other receiving sound at various locations. Analyzing this sound helps to inversely characterize the acoustic primitives. Our framework is well-grounded in the fundamental physical principles of sound propagation, including reciprocity and globality, and thus is acoustically interpretable and meaningful. We present experiments on both synthetic and real-world datasets, demonstrating superior quality in RIR estimation against closely related methods.

Automating the assembly of objects from their parts is a complex problem with innumerable applications in manufacturing, maintenance, and recycling. Unlike existing research, which is limited to target segmentation, pose regression, or using fixed target blueprints, our work presents a holistic multi-level framework for part assembly planning consisting of part assembly sequence inference, part motion planning, and robot contact optimization. We present the Part Assembly Sequence Transformer (PAST) – a sequence-to-sequence neural network – to infer assembly sequences recursively from a target blueprint. We then use a motion planner and optimization to generate part movements and contacts. To train PAST, we introduce D4PAS: a large-scale Dataset for Part Assembly Sequences consisting of physically valid sequences for industrial objects. Experimental results show that our approach generalizes better than prior methods while needing significantly less computational time for inference. Further details on our experiments and results are available in the video.

In this paper, we consider the problem of segmenting multiple instances of a transparent object from RGB or gray scale camera images in a robotic bin picking setting. Prior methods for solving this task are usually built on the Mask-RCNN framework, but they require large annotated datasets for fine-tuning. Instead, we consider the task in a few-shot setting and present TrInSeg, a data-efficient and robust instance segmentation method for transparent objects based on Mask-RCNN. Our key innovations in TrInSeg are twofold: i) a novel method, dubbed TransMixup, for producing new training images using synthetic transparent object instances created by spatially transforming annotated examples; and ii) a method for scoring the consistency between the predicted segments and rotations of an ideal object template. In our new scoring method, the spatial transformations are produced by an auxiliary neural network, and the scores are then used to filter inconsistent instance predictions. To demonstrate the effectiveness of our method, we present experiments on a new few-shot dataset consisting of seven categories of non-opaque (transparent and translucent) objects, each category varying in the size, shape, and degree of transparency of the objects. Our results show that TrInSeg achieves state-of-the-art performance, improving fine-tuned Mask-RCNN by more than 14% in mIoU, while requiring very few annotated training samples.

We study the problem of multimodal physical scene understanding, where an embodied agent needs to find fallen objects by inferring object properties, direction, and distance of an impact sound source. Previous works adopt feed-forward neural networks to directly regress the variables from sound, leading to poor generalization and domain adaptation issues. In this paper, we illustrate that learning a disentangled model of acoustic formation, referred to as disentangled acoustic field (DAF), to capture the sound generation and propagation process, enables the embodied agent to construct a spatial uncertainty map over where the objects may have fallen. We demonstrate that our analysis-by-synthesis framework can jointly infer sound properties by explicitly decomposing and factorizing the latent space of the disentangled model. We further show that the spatial uncertainty map can significantly improve the success rate for the localization of fallen objects by proposing multiple plausible exploration locations.

Recent years have seen a significant progress in the general-purpose problem solving abilities of large vision and language models (LVLMs), such as ChatGPT, Gemini, etc.; some of these breakthroughs even seem to enable AI models to outperform human abilities in varied tasks that demand higher-order cognitive skills. Are the current large AI models indeed capable of generalized problem solving as humans do? A systematic analysis of AI capabilities for joint vision and text reasoning, however, is missing in the current scientific literature. In this paper, we make an effort towards filling this gap, by evaluating state-of-the-art LVLMs on their mathematical and algorithmic reasoning abilities using visuo-linguistic problems from children's Olympiads. Specifically, we consider problems from the Mathematical Kangaroo (MK) Olympiad, which is a popular international competition targeted at children from grades 1-12, that tests children's deeper mathematical abilities using puzzles that are appropriately gauged to their age and skills. Using the puzzles from MK, we created a dataset, dubbed SMART-840, consisting of 840 problems from years 2020-2024. With our dataset, we analyze LVLMs power on mathematical reasoning; their responses on our puzzles offer a direct way to compare against that of children. Our results show that modern LVLMs do demonstrate increasingly powerful reasoning skills in solving problems for higher grades, but lack the foundations to correctly answer problems designed for younger children. Further analysis shows that there is no significant correlation between the reasoning capabilities of AI models and that of young children, and their capabilities appear to be based on a different type of reasoning than the cumulative knowledge that underlies children's mathematical skills.

Prototypical part neural networks (ProtoPartNNs), namely ProtoPNet and its derivatives, are an intrinsically interpretable approach to machine learning. Their prototype learning scheme enables intuitive explanations of the form, this (prototype) looks like that (testing image patch). But, does this actually look like that? In this work, we delve into why object part localization and associated heat maps in past work are misleading. Rather than localizing to object parts, existing ProtoPartNNs localize to the entire image, contrary to generated explanatory visualizations. We argue that detraction from these underlying issues is due to the alluring nature of visualizations and an over-reliance on intuition. To alleviate these issues, we devise new receptive field-based architectural constraints for meaningful localization and a principled pixel space mapping for ProtoPartNNs. To improve interpretability, we propose additional architectural improvements, including a simplified classification head. We also make additional corrections to ProtoPNet and its derivatives, such as the use of a validation set, rather than a test set, to evaluate generalization during training. Our approach, PixPNet (Pixel-grounded Prototypical part Network), is the only ProtoPartNN that truly learns and localizes to prototypical object parts. We demonstrate that PixPNet achieves quantifiably improved interpretability without sacrificing accuracy1.

Spatial localization of 3D sound sources is an important problem in many real world scenarios, especially when the sources may not have any visually distinguishable characteristic; e.g., finding a gas leak, a malfunctioning motor, etc. In this paper, we cast this task in a novel audio-visual setting, by introducing an acoustic-camera rig consisting of a centered pinhole RGB camera and a uniform circular array of four coplanar microphones. Using this setup, we propose Sound3DVDet – a 3D sound source localization Transformer model that treats this task as a set prediction problem. It first learns a set of initial sound source locations (dubbed queries) from a single view of the microphone array signal, then feeds the query set to a sequence of Transformerlike layers for refinement. Each query arising from each layer repeatedly aggregates sound source cues from other views. We deeply supervise the initial sound source queries, intermediate layer queries, and the final output by measuring their respective discrepancy against ground truth queries via bipartite matching. To evaluate our method, we introduce a new dataset: Sound3DVDet Dataset, consisting of nearly 6k scenes produced using the SoundSpaces simulator. We conduct extensive experiments on our dataset and show the efficacy of our approach against closely related methods, demonstrating significant improvements in the localization accuracy. Code is available at https://github.com/yuhanghe01/Sound3DVDet.

Recent times have witnessed an increasing number of applications of deep neural networks towards solving tasks that require superior cognitive abilities, e.g., playing Go, generating art, question answering (e.g., ChatGPT), etc. Such a dramatic progress raises the question: how generalizable are neural networks in solving problems that demand broad skills? To answer this question, we propose SMART: a Simple Multimodal Algorithmic Reasoning Task and the associated SMART-101 dataset11The SMART-101 dataset is publicly available at: https://doi.org/10.5281/zenodo.7761800, for evaluating the abstraction, deduction, and generalization abilities of neural networks in solving visuo-linguistic puzzles designed specifically for children in the 6–8 age group. Our dataset consists of 101 unique puzzles; each puzzle comprises a picture and a question, and their solution needs a mix of several elementary skills, including arithmetic, algebra, and spatial reasoning, among others. To scale our dataset towards training deep neural networks, we programmatically generate entirely new instances for each puzzle while retaining their solution algorithm. To benchmark the performance on the SMART-101 dataset, we propose a vision-and-language meta-learning model that can incorporate varied state-of-the-art neural backbones. Our experiments reveal that while powerful deep models offer reasonable performances on puzzles in a supervised setting, they are not better than random accuracy when analyzed for generalization –filling this gap may demand new multimodal learning approaches.

Supervised learning of skeleton sequence encoders for action recognition has received significant attention in recent times. However, learning such encoders without labels continues to be a challenging problem. While prior works have shown promising results by applying contrastive learning to pose sequences, the quality of the learned representations is often observed to be closely tied to data augmentations that are used to craft the positives. However, augmenting pose sequences is a difficult task as the geometric constraints among the skeleton joints need to be enforced to make the augmentations realistic for that action. In this work, we propose a new contrastive learning approach to train models for skeleton-based action recognition without labels. Our key contribution is a simple module, HaLP - to Hallucinate Latent Positives for contrastive learning. Specifically, HaLP explores the latent space of poses in suitable directions to generate new positives. To this end, we present a novel optimization formulation to solve for the synthetic positives with an explicit control on their hardness. We propose approximations to the objective, making them solvable in closed form with minimal overhead. We show via experiments that using these generated positives within a standard contrastive learning framework leads to consistent improvements across benchmarks such as NTU-60, NTU-120, and PKU-II on tasks like linear evaluation, transfer learning, and kNN evaluation. Our code can be found at https://github.com/anshulbshah/HaLP.

Multimodal alignment facilitates the retrieval of instances from one modality when queried using another. In this paper, we consider a novel setting where such an alignment is between (i) instruction steps that are depicted as assembly diagrams (commonly seen in Ikea assembly manuals) and (ii) segments from in-the-wild videos; these videos comprising an enactment of the assembly actions in the real world. We introduce a supervised contrastive learning approach that learns to align videos with the subtle details of assembly diagrams, guided by a set of novel losses. To study this problem and evaluate the effectiveness of our method, we introduce a new dataset: IAW-for Ikea assembly in the wild-consisting of 183 hours of videos from diverse furniture assembly collections and nearly 8,300 illustrations from their associated instruction manuals and annotated for their ground truth alignments. We define two tasks on this dataset: First, nearest neighbor retrieval between video segments and illustrations, and, second, alignment of instruction steps and the segments for each video. Extensive experiments on IAW demonstrate superior performance of our approach against alternatives.

Conditional generative models typically demand large annotated training sets to achieve high-quality synthesis. As a result, there has been significant interest in designing models that perform plug-and-play generation, i.e., to use a predefined or pretrained model, which is not explicitly trained on the generative task, to guide the generative process (e.g., using language). However, such guidance is typically useful only towards synthesizing high-level semantics rather than editing fine-grained details as in image-to-image translation tasks. To this end, and capitalizing on the powerful fine-grained generative control offered by the recent diffusion-based generative models, we introduce Steered Diffusion, a generalized framework for photorealistic zero-shot conditional image generation using a diffusion model trained for unconditional generation. The key idea is to steer the image generation of the diffusion model at inference time via designing a loss using a pre-trained inverse model that characterizes the conditional task. This loss modulates the sampling trajectory of the diffusion process. Our framework allows for easy incorporation of multiple conditions during inference. We present experiments using steered diffusion on several tasks including inpainting, colorization, text-guided semantic editing, and image super-resolution. Our results demonstrate clear qualitative and quantitative improvements over state-of-the-art diffusion-based plug-and-play models while adding negligible additional computational cost.

In this paper, we present a simple and efficient scheme for segmenting approximately convex 3D object instances in depth images in a few-shot setting via discriminatively modeling the 3D shape of the object using a neural network. Our key idea is to select pairs of 3D points on the depth image between which we compute surface geodesics. As the number of such geodesics is quadratic in the number of image pixels, we can create a large training set of geodesics using only very limited ground truth instance annotations. These annotations are used to create a binary label for each geodesic, which indicates whether or not that geodesic belongs entirely to one instance segment. A neural network is then trained to classify the geodesics using these labels. During inference, we create geodesics from selected seed points in the test depth image, then produce a convex hull of the points that are classified by the neural network as belonging to the same instance, thereby achieving instance segmentation. We present experiments applying our method to segmenting instances of food items in real-world depth images. Our results demonstrate promising performances compared to prior methods in accuracy and computational efficiency.

The world is filled with articulated objects that are difficult to determine how to use from vision alone, e.g., a door might open inwards or outwards. Humans handle these objects with strategic trial-and-error: first pushing a door then pulling if that doesn't work. We enable these capabilities in autonomous agents by proposing “Hypothesize, Simulate, Act, Update, and Repeat” (H-SAUR), a probabilistic generative framework that simultaneously generates a distribution of hypotheses about how objects articulate given input observations, captures certainty over hypotheses over time, and infer plausible actions for exploration and goal-conditioned manipulation. We compare our model with existing work in manipulating objects after a handful of exploration actions, on the PartNet-Mobility dataset. We further propose a novel PuzzleBoxes benchmark that contains locked boxes that require multiple steps to solve. We show that the proposed model significantly outperforms the current state-of-the-art articulated object manipulation framework, despite using zero training data. We further improve the test-time efficiency of H-SAUR by integrating a learned prior from learning-based vision models.

Spatio-temporal scene-graph approaches to video-based reasoning tasks, such as video question-answering (QA), typically construct such graphs for every video frame. These approaches often ignore the fact that videos are essentially sequences of 2D ``views'' of events happening in a 3D space, and that the semantics of the 3D scene can thus be carried over from frame to frame. Leveraging this insight, we propose a (2.5+1)D scene graph representation to better capture the spatio-temporal information flows inside the videos. Specifically, we first create a 2.5D (pseudo-3D) scene graph by transforming every 2D frame to have an inferred 3D structure using an off-the-shelf 2D-to-3D transformation module, following which we register the video frames into a shared (2.5+1)D spatio-temporal space and ground each 2D scene graph within it. Such a (2.5+1)D graph is then segregated into a static sub-graph and a dynamic sub-graph, corresponding to whether the objects within them usually move in the world. The nodes in the dynamic graph are enriched with motion features capturing their interactions with other graph nodes. Next, for the video QA task, we present a novel transformer-based reasoning pipeline that embeds the (2.5+1)D graph into a spatio-temporal hierarchical latent space, where the sub-graphs and their interactions are captured at varied granularity. To demonstrate the effectiveness of our approach, we present experiments on the NExT-QA and AVSD-QA datasets. Our results show that our proposed (2.5+1)D representation leads to faster training and inference, while our hierarchical model showcases superior performance on the video QA task versus the state of the art.

Recent advances in generative adversarial networks (GANs) have led to remarkable achievements in face image synthesis. While methods that use style-based GANs can generate strikingly photorealistic face images, it is often difficult to control the characteristics of the generated faces in a meaningful and disentangled way. Prior approaches aim to achieve such semantic control and disentanglement within the latent space of a previously trained GAN. In contrast, we propose a framework that a priori models physical attributes of the face such as 3D shape, albedo, pose, and lighting explicitly, thus providing disentanglement by design. Our method, MOST-GAN, integrates the expressive power and photorealism of style-based GANs with the physical disentanglement and flexibility of nonlinear 3D morphable models, which we couple with a state-of-the-art 2D hair manipulation network. MOST-GAN achieves photorealistic manipulation of portrait images with fully disentangled 3D control over their physical attributes, enabling extreme manipulation of lighting, facial expression, and pose variations up to full profile view.

Standard contrastive learning approaches usually require a large number of negatives for effective unsupervised learning and often exhibit slow convergence. We suspect this behavior is due to the suboptimal selection of negatives used for offering contrast to the positives. We counter this difficulty by taking inspiration from support vector machines (SVMs) to present max-margin contrastive learning (MMCL). Our approach selects negatives as the sparse support vectors obtained via a quadratic optimization problem, and contrastiveness is enforced by maximizing the decision margin. As SVM optimization can be computationally demanding, especially in an end-to-end setting, we present simplifications that alleviate the computational burden. We validate our approach on standard vision benchmark datasets, demonstrating better performance in unsupervised representation learning over state-of-the-art, while having better empirical convergence properties.

In previous work, we have proposed the Audio-Visual Scene-Aware Dialog (AVSD) task, collected an AVSD dataset, developed AVSD technologies, and hosted an AVSD challenge track at both the 7th and 8th Dialog System Technology Challenges (DSTC7, DSTC8). In these challenges, the best-performing systems relied heavily on human-generated descriptions of the video content, which were available in the datasets but would be unavailable in real-world applications. To promote further advancements for real-world applications, we proposed a third AVSD challenge, at DSTC10, with two modifications: 1) the human-created description is unavailable at inference time, and 2) systems must demonstrate temporal reasoning by finding evidence from the video to support each answer. This paper introduces the new task that includes temporal reasoning and our new extension of the AVSD dataset for DSTC10, for which we collected human-generated temporal reasoning data. We also introduce a baseline system built using an AV-transformer, which we released along with the new dataset. Finally, this paper introduces a new system that extends our baseline system with attentional multimodal fusion, joint student-teacher learning (JSTL), and model combination techniques, achieving state-of-the-art performances on the AVSD datasets for DSTC7, DSTC8, and DSTC10. We also propose two temporal reasoning methods for AVSD: one attention-based, and one based on a time-domain region proposal network.

Recent years have seen embodied visual navigation advance in two distinct directions: (i) in equipping the AI agent to follow natural language instructions, and (ii) in making the navigable world multimodal, e.g., audio-visual navigation. However, the real world is not only multimodal, but also often complex, and thus in spite of these advances, agents still need to understand the uncertainty in their actions and seek instructions to navigate. To this end, we present AVLEN -- an interactive agent for Audio-Visual-Language Embodied Navigation. Similar to audio-visual navigation tasks, the goal of our embodied agent is to localize an audio event via navigating the 3D visual world; however, the agent may also seek help from a human (oracle), where the assistance is provided in free-form natural language. To realize these abilities, AVLEN uses a multimodal hierarchical reinforcement learning backbone that learns: (a) high-level policies to choose either audio-cues for navigation or to query the oracle, and (b) lower-level policies to select navigation actions based on its audio-visual and language inputs. The policies are trained via rewarding for the success on the navigation task while minimizing the number of queries to the oracle. To empirically evaluate AVLEN, we present experiments on the SoundSpaces framework for semantic audio-visual navigation tasks. Our results show that equipping the agent to ask for help leads to a clear improvement in performances, especially in challenging cases, e.g., when the sound is unheard during training or in the presence of distractor sounds.

There exists an unequivocal distinction between the sound produced by a static source and that produced by a moving one, especially when the source moves towards or away from the microphone. In this paper, we propose to use this connection between audio and visual dynamics for solving two challenging tasks simultaneously, namely: (i) separating audio sources from a mixture using visual cues, and (ii) predicting the 3D visual motion of a sounding source using its separated audio. Towards this end, we present Audio Separator and Motion Predictor (ASMP) -- a deep learning framework that leverages the 3D structure of the scene and the motion of sound sources for better audio source separation. At the heart of ASMP is a 2.5D scene graph capturing various objects in the video and their pseudo-3D spatial proximities. This graph is constructed by registering together 2.5D monocular depth predictions from the 2D video frames and associating the 2.5D scene regions with the outputs of an object detector applied on those frames. The ASMP task is then mathematically modeled as the joint problem of: (i) recursively segmenting the 2.5D scene graph into several sub-graphs, each associated with a constituent sound in the input audio mixture (which is then separated) and (ii) predicting the 3D motions of the corresponding sound sources from the separated audio. To empirically evaluate ASMP, we present experiments on two challenging audio-visual datasets, viz. Audio Separation in the Wild (ASIW) and Audio Visual Event (AVE). Our results demonstrate that ASMP achieves a clear improvement in source separation quality, outperforming prior works on both datasets, while also estimating the direction of motion of the sound sources better than other methods.

Learning from a few examples is a challenging computer vision task. Traditionally,meta-learning-based methods have shown promise towards solving this problem.Recent approaches show benefits by learning a feature extractor on the abundantbase examples and transferring these to the fewer novel examples. However, thefinetuning stage is often prone to overfitting due to the small size of the noveldataset. To this end, we propose Few shot Learning with hard Mixup (FeLMi)using manifold mixup to synthetically generate samples that helps in mitigatingthe data scarcity issue. Different from a naïve mixup, our approach selects the hardmixup samples using an uncertainty-based criteria. To the best of our knowledge,we are the first to use hard-mixup for the few-shot learning problem. Our approachallows better use of the pseudo-labeled base examples through base-novel mixupand entropy-based filtering. We evaluate our approach on several common few-shotbenchmarks - FC-100, CIFAR-FS, miniImageNet and tieredImageNet and obtainimprovements in both 1-shot and 5-shot settings. Additionally, we experimented onthe cross-domain few-shot setting (miniImageNet → CUB) and obtain significantimprovements.

Representations in the form of Symmetric Positive Definite (SPD) matrices have been popularized in a variety of visual learning applications due to their demonstrated ability to capture rich second-order statistics of visual data. There exist several similarity measures for comparing SPD matrices with documented benefits. However, selecting an appropriate measure for a given problem remains a challenge and in most cases, is the result of a trial-and-error process. In this paper, we propose to learn similarity measures in a data-driven manner. To this end, we capitalize on the αβ-log-det divergence, which is a meta-divergence parametrized by scalars α and β, subsuming a wide family of popular information divergences on SPD matrices for distinct and discrete values of these parameters. Our key idea is to cast these parameters in a continuum and learn them from data. We systematically extend this idea to learn vector-valued parameters, thereby increasing the expressiveness of the underlying non-linear measure. We conjoin the divergence learning problem with several standard tasks in machine learning, including supervised discriminative dictionary learning and unsupervised SPD matrix clustering. We present Riemannian gradient descent schemes for optimizing our formulations efficiently, and show the usefulness of our method on eight standard computer vision tasks.

One-class learning is the classic problem of fitting a model to the data for which annotations are available only for a single class. In this paper, we explore novel objectives for one-class learning, which we collectively refer to as Generalized One-class Discriminative Subspaces (GODS). Our key idea is to learn a pair of complementary classifiers to flexibly bound the one-class data distribution, where the data belongs to the positive half-space of one of the classifiers in the complementary pair and to the negative half-space of the other. To avoid redundancy while allowing non-linearity in the classifier decision surfaces, we propose to design each classifier as an orthonormal frame and seek to learn these frames via jointly optimizing for two conflicting objectives, namely: i) to minimize the distance between the two frames, and ii) to maximize the margin between the frames and the data. The learned orthonormal frames will thus characterize a piecewise linear decision surface that allows for efficient inference, while our objectives seek to bound the data within a minimal volume that maximizes the decision margin, thereby robustly capturing the data distribution. We explore several variants of our formulation under different constraints on the constituent classifiers, including kernelized feature maps. We demonstrate the empirical benefits of our approach via experiments on data from several applications in computer vision, such as anomaly detection in video sequences, human poses, and human activities. We also explore the generality and effectiveness of GODS for non-vision tasks via experiments on several UCI datasets, demonstrating state-of-the-art results.

Human actions in video sequences are characterized by the complex interplay between spatial features and their temporal dynamics. In this paper, we propose novel tensor representations for compactly capturing such higher-order relationships between visual features for the task of action recognition. We propose two tensor-based feature representations, viz. (i)sequence compatibility kernel(SCK) and (ii)dynamics compatibility kernel(DCK). SCK builds on the spatio-temporal correlations between features, whereas DCK explicitly models the action dynamics of a sequence. We also explore generalization of SCK, coined SCK$\;\oplus$, that operates on subsequences to capture the local-global interplay of correlations, which can incorporate multi-modal inputs e.g., skeleton 3D body-joints and per-frame classifier scores obtained from deep learning models trained on videos. We introduce linearization of these kernels that lead to compact and fast descriptors. We provide experiments on (i) 3D skeleton action sequences, (ii) fine-grained video sequences, and (iii) standard non-fine-grained videos. As our final representations are tensors that capture higher-order relationships of features, they relate to co-occurrences for robust fine-grained recognition (Lin, 2017), (Koniusz, 2018). We use higher-order tensors and so-called Eigenvalue Power Normalization (EPN) which have been long speculated to perform spectral detection of higher-order occurrences (Koniusz, 2013), (Koniusz, 2017), thus detecting fine-grained relationships of features rather than merely count features in action sequences. We prove that a tensor of order$r$, built from$Z_*$dimensional features, coupled with EPN indeed detects if at least one higher-order occurrence is ‘projected’ into one of its$\binom{Z_*}{r}$subspaces of dim.$r$represented by the tensor, thus forming a Tensor Power Normalization metric endowed with$\binom{Z_*}{r}$such ‘detectors’.

Given an input video, its associated audio, and a brief caption, the audio-visual scene aware dialog (AVSD) task requires an agent to indulge in a question-answer dialog with a human about the audio-visual content. This task thus poses a challenging multi-modal representation learning and reasoning scenario, advancements into which could influence several human-machine interaction applications. To solve this task, we introduce a semantics-controlled multi-modal shuffled Transformer reasoning framework, consisting of a sequence of Transformer modules, each taking a modality as input and producing representations conditioned on the input question. Our proposed Transformer variant uses a shuffling scheme on their multi-head outputs, demonstrating better regularization. To encode fine-grained visual information, we present a novel dynamic scene graph representation learning pipeline that consists of an intra-frame reasoning layer producing spatio-semantic graph representations for every frame, and an inter-frame aggregation module capturing temporal cues. Our entire pipeline is trained end-to-end. We present experiments on the benchmark AVSD dataset, both on answer generation and selection tasks. Our results demonstrate state-of-the-art performances on all evaluation metrics.

Predicting the future frames of a video is a challenging task, in part due to the underlying stochastic real-world phenomena. Prior approaches to solve this task typically estimate a latent prior characterizing this stochasticity, however do not account for the predictive uncertainty of the (deep learning) model. Such approaches often derive the training signal from the mean-squared error (MSE) between the generated frame and the ground truth, which can lead to sub-optimal training, especially when the predictive uncertainty is high. Towards this end, we introduce Neural Uncertainty Quantifier (NUQ) - a stochastic quantification of the model’s predictive uncertainty, and use it to weigh the MSE loss. We propose a hierarchical, variational framework to derive NUQ in a principled manner using a deep, Bayesian graphical model. Our experiments on three benchmark stochastic video prediction datasets show that our proposed framework trains more effectively compared to the state-of-the-art models (especially when the training sets are small), while demonstrating better video generation quality and diversity against several evaluation metrics.

State-of-the-art approaches for visually-guided audio source separation typically assume sources that have characteristic sounds, such as musical instruments. These approaches often ignore the visual context of these sound sources or avoid modeling object interactions that may be useful to better characterize the sources, especially when the same object class may produce varied sounds from distinct interactions. To address this challenging problem, we propose Audio Visual Scene Graph Segmenter (AVSGS), a novel deep learning model that embeds the visual structure of the scene as a graph and segments this graph into subgraphs, each subgraph being associated with a unique sound obtained by co-segmenting the audio spectrogram. At its core, AVSGS uses a recursive neural network that emits mutually-orthogonal sub-graph embeddings of the visual graph using multi-head attention. These embeddings are used for conditioning an audio encoder-decoder towards source separation. Our pipeline is trained end-to-end via a self-supervised task consisting of separating audio sources using the visual graph from artificially mixed sounds.In this paper, we also introduce an “in the wild” video dataset for sound source separation that contains multiple non-musical sources, which we call Audio Separation in the Wild (ASIW). This dataset is adapted from the AudioCaps dataset, and provides a challenging, natural, and daily-life setting for source separation. Thorough experiments on the proposed ASIW and the standard MUSIC datasets demonstrate state-of-the-art sound separation performance of our method against recent prior approaches.

In this paper, we present InSeGAN, an unsupervised 3D generative adversarial network (GAN) for segmenting (nearly) identical instances of rigid objects in depth images. Using an analysis-by-synthesis approach, we design a novel GAN architecture to synthesize a multiple-instance depth image with independent control over each instance. InSeGAN takes in a set of code vectors (e.g., random noise vectors), each encoding the 3D pose of an object that is represented by a learned implicit object template. The generator has two distinct modules. The first module, the instance feature generator, uses each encoded pose to transform the implicit template into a feature map representation of each object instance. The second module, the depth image renderer, aggregates all of the single-instance feature maps output by the first module and generates a multiple-instance depth image. A discriminator distinguishes the generated multiple-instance depth images from the distribution of true depth images. To use our model for instance segmentation, we propose an instance pose encoder that learns to take in a generated depth image and reproduce the pose code vectors for all of the object instances. To evaluate our approach, we introduce a new synthetic dataset, "Insta-10," consisting of 100,000 depth images, each with 5 instances of an object from one of 10 classes. Our experiments on Insta-10, as well as on real-world noisy depth images, show that InSeGAN achieves state-of-the-art performance, often outperforming prior methods by large margins.

Most popular deep models for action recognition in videos generate independent predictions for short clips, which are then pooled heuristically to assign an action label to the full video segment. As not all frames may characterize the underlying action-indeed, many are common across multiple actions-pooling schemes that impose equal importance on all frames might be unfavorable. In an attempt to tackle this problem, we propose discriminative pooling, based on the notion that among the deep features generated on all short clips, there is at least one that characterizes the action. To identify these useful features, we resort to a negative bag consisting of features that are known to be irrelevant, for example, they are sampled either from datasets that are unrelated to our actions of interest or are CNN features produced via random noise as input. With the features from the video as a positive bag and the irrelevant features as the negative bag, we cast an objective to learn a (nonlinear) hyperplane that separates the unknown useful features from the rest in a multiple instance learning formulation within a support vector machine setup. We use the parameters of this separating hyperplane as a descriptor for the full video segment. Since these parameters are directly related to the support vectors in a max-margin framework, they can be treated as a weighted average pooling of the features from the bags, with zero weights given to non-support vectors. Our pooling scheme is end-to-end trainable within a deep learning framework. We report results from experiments on eight computer vision benchmark datasets spanning a variety of video-related tasks and demonstrate state-of-the-art performance across these tasks.

This paper introduces the Eighth Dialog System Technology Challenge. In line with recent challenges, the eighth edition focuses on applying end-to-end dialog technologies in a pragmatic way for multi-domain task-completion, noetic response selection, audio visual scene-aware dialog, and schema-guided dialog state tracking tasks. This paper describes the task definition, provided datasets, baselines and evaluation set-up for each track. We also summarize the results of the submitted systems to highlight the overall trends of the state-of-the-art technologies for the tasks.

Modern face alignment methods have become quite accurate at predicting the locations of facial landmarks, but they do not typically estimate the uncertainty of their predicted locations nor predict whether landmarks are visible. In this paper, we present a novel framework for jointly predicting landmark locations, associated uncertainties of these predicted locations, and landmark visibilities. We model these as mixed random variables and estimate them using a deep network trained using our proposed Location, Uncertainty, and Visibility Likelihood (LUVLi) loss. In addition, we release an entirely new labeling of a large face alignment dataset with over 19,000 face images in a full range of head poses. Each face is manually labeled with the ground-truth locations of 68 landmarks, with the additional information of whether each landmarks is visible, self-occluded (due to extreme head poses), or externally occluded. Not only does our joint estimation yield accurate estimates of the uncertainty of predicted landmark locations, but it also yields state-of-the-art estimates for the landmark locations themselves on mulitple standard face alignment datasets. Our method's estimates of the uncertainty of predicted landmark locations could be used to automatically identify input images on which face alignment fails, which can be critical for downstream tasks.

In this paper, we study the problem of learning compact (low-dimensional) representations for sequential data that captures its implicit spatio-temporal cues. To maximize extraction of such informative cues from the data, we set the problem within the context of contrastive representation learning and to that end propose a novel objective via optimal transport. Specifically, our formulation seeks a low-dimensional subspace representation of the data that jointly (i) maximizes the distance of the data (embedded in this subspace) from an adversarial data distribution under the optimal transport, a.k.a. the Wasserstein distance, (ii) captures the temporal order, and (iii) minimizes the data distortion. To generate the adversarial distribution, we propose a novel framework connecting Wasserstein GANs with a classifier, allowing a principled mechanism for producing good negative distributions for contrastive learning, which is currently a challenging problem. Our full objective is cast as a subspace learning problem on the Grassmann manifold and solved via Riemannian optimization. To empirically study our formulation, we provide experiments on the task of human action recognition in video sequences. Our results demonstrate competitive performance against challenging baselines.

Generating video descriptions automatically is a challenging task that involves a complex interplay between spatio-temporal visual features and language models. Given that videos consist of spatial (frame-level) features and their temporal evolutions, an effective captioning model should be able to attend to these different cues selectively. To this end, we propose a Spatio-Temporal and Temporo-Spatial (STaTS) attention model which, conditioned on the language state, hierarchically combines spatial and temporal attention to videos in two different orders: (i) a spatiotemporal (ST) sub-model, which first attends to regions that have temporal evolution, then temporally pools the features from these regions; and (ii) a temporo-spatial (TS) sub-model, which first decides a single frame to attend to, then applies spatial attention within that frame. We propose a novel LSTM-based temporal ranking function, which we call ranked attention, for the ST model to capture action dynamics. Our entire framework is trained end-to-end. We provide experiments on two benchmark datasets: MSVD and MSR-VTT. Our results demonstrate the synergy between the ST and TS modules, outperforming recent state-of-the-art methods.

We propose a self-supervised approach to improve the training of Generative Adversarial Networks (GANs) via inducing the discriminator to examine the structural consistency of images. Although natural image samples provide ideal examples of both valid structure and valid texture, learning to reproduce both together remains an open challenge. In our approach, we augment the training set of natural images with modified examples that have degraded structural consistency. These degraded examples are automatically created by randomly exchanging pairs of patches in an image’s convolutional feature map. We call this approach feature exchange. With this setup, we propose a novel GAN formulation, termed Feature eXchange GAN (FX-GAN), in which the discriminator is trained not only to distinguish real versus generated images, but also to perform the auxiliary task of distinguishing between real images and structurally corrupted (feature-exchanged) real images. This auxiliary task causes the discriminator to learn the proper feature structure of natural images, which in turn guides the generator to produce images with more realistic structure. Compared with strong GAN baselines, our proposed self-supervision approach improves generated image quality, diversity, and training stability for both the unconditional and class-conditional settings.

We introduce the task of scene-aware dialog. Our goal is to generate a complete and natural response to a question about a scene, given video and audio of the scene and the history of previous turns in the dialog. To answer successfully, agents must ground concepts from the question in the video while leveraging contextual cues from the dialog history. To benchmark this task, we introduce the Audio Visual Scene-Aware Dialog (AVSD) Dataset. For each of more than 11,000 videos of human actions from the Charades dataset, our dataset contains a dialog about the video, plus a final summary of the video by one of the dialog participants. We train several baseline systems for this task and evaluate the performance of the trained models using both qualitative and quantitative metrics. Our results indicate that models must utilize all the available inputs (video, audio, question, and dialog history) to perform best on this dataset.

In order for machines interacting with the real world to have conversations with users about the objects and events around them, they need to understand dynamic audiovisual scenes. The recent revolution of neural network models allows us to combine various modules into a single end-to-end differentiable network. As a result, Audio Visual Scene-Aware Dialog (AVSD) systems for real-world applications can be developed by integrating state-of-the-art technologies from multiple research areas, including end-to-end dialog technologies, visual question answering (VQA) technologies, and video description technologies. In this paper, we introduce a new data set of dialogs about videos of human behaviors, as well as an end-to-end Audio Visual Scene-Aware Dialog (AVSD) model, trained using this new data set, that generates responses in a dialog about a video. By using features that were developed for multimodal attention-based video description, our system improves the quality of generated dialog about dynamic video scenes.

One-class learning is the classic problem of fitting a model to data for which annotations are available only for a single class. In this paper, we propose a novel objective for one-class learning. Our key idea is to use a pair of orthonormal frames - as subspaces - to "sandwich'' the labeled data via optimizing for two objectives jointly: i) minimize the distance between the origins of the two subspaces, and ii) to maximize the margin between the hyperplanes and the data, either subspace demanding the data to be in its positive and negative orthant respectively. Our proposed objective however leads to a non-convex optimization problem, to which we resort to Riemannian optimization schemes and derive an efficient conjugate gradient scheme on the Stiefel manifold. To study the effectiveness of our scheme, we propose a new dataset Dash-Cam-Pose, consisting of clips with skeleton poses of humans seated in a car, the task being to classify the clips as normal or abnormal; the latter is when any human pose is out-of-position with regard to say an airbag deployment. Our experiments on the proposed Dash-Cam-Pose dataset, as well as several other standard anomaly/novelty detection benchmarks demonstrate the benefits of our scheme, achieving state-of-the-art one-class accuracy.

Computing Nash equilibrium (NE) of multi-player games has witnessed renewed interest due to recent advances in generative adversarial networks. However, computing equilibrium efficiently is challenging. To this end, we introduce the Gradient-based Nikaido-Isoda (GNI) function which serves: (i) as a merit function, vanishing only at the first-order stationary points of each player’s optimization problem, and (ii) provides error bounds to a stationary Nash point. Gradient descent is shown to converge sublinearly to a first-order stationary point of the GNI function. For the particular case of bilinear min-max games and multi-player quadratic games, the GNI function is convex. Hence, the application of gradient descent in this case yields linear convergence to an NE (when one exists). In our numerical experiments, we observe that the GNI formulation always converges to the first-order stationary point of each player’s optimization problem.

Unpaired image-to-image translation is the problem of mapping an image in the source domain to one in the target domain, without requiring corresponding image pairs. To ensure the translated images are realistically plausible, recent works, such as Cycle-GAN, demands this mapping to be invertible. While, this requirement demonstrates promising results when the domains are unimodal, its performance is unpredictable in a multi-modal scenario such as in an image segmentation task. This is because, invertibility does not necessarily enforce semantic correctness. To this end, we present a semantically-consistent GAN framework, dubbed Sem-GAN, in which the semantics are defined by the class identities of image segments in the source domain as produced by a semantic segmentation algorithm. Our proposed framework includes consistency constraints on the translation task that, together with the GAN loss and the cycle-constraints, enforces that the images when translated will inherit the appearances of the target domain, while (approximately) maintaining their identities from the source domain. We present experiments on several image-to-image translation tasks and demonstrate that Sem-GAN improves the quality of the translated images significantly, sometimes by more than 20% on the FCN score. Further, we show that semantic segmentation models, trained with synthetic images translated via Sem-GAN, leads to significantly better segmentation results than other variants.

We present a principled approach to uncover the structure of visual data by solving a deep learning task coined visual permutation learning. The goal of this task is to find the permutation that recovers the structure of data from shuffled versions of it. In the case of natural images, this task boils down to recovering the original image from patches shuffled by an unknown permutation matrix. Permutation matrices are discrete, thereby posing difficulties for gradient-based optimization methods. To this end, we resort to a continuous approximation using doubly-stochastic matrices and formulate a novel bi-level optimization problem on such matrices that learns to recover the permutation. Unfortunately, such a scheme leads to expensive gradient computations. We circumvent this issue by further proposing a computationally cheap scheme for generating doubly stochastic matrices based on Sinkhorn iterations. To implement our approach we propose DeepPermNet, an end-to-end CNN model for this task. The utility of DeepPermNet is demonstrated on three challenging computer vision problems, namely, relative attributes learning, supervised learning-to-rank, and self-supervised representation learning. Our results show state-of-the-art performance on the Public Figures and OSR benchmarks for relative attributes learning, chronological and interestingness image ranking for supervised learning-to-rank, and competitive results in the classification and segmentation tasks of the PASCAL VOC dataset for self-supervised representation learning.

Representations that can compactly and effectively capture the temporal evolution of semantic content are important to computer vision and machine learning algorithms that operate on multi-variate time-series data. We investigate such representations motivated by the task of human action recognition. Here each data instance is encoded by a multivariate feature (such as via a deep CNN) where action dynamics are characterized by their variations in time. As these features are often non-linear, we propose a novel pooling method, kernelized rank pooling, that represents a given sequence compactly as the pre-image of the parameters of a hyperplane in a reproducing kernel Hilbert space, projections of data onto which captures their temporal order. We develop this idea further and show that such a pooling scheme can be cast as an order-constrained kernelized PCA objective. We then propose to use the parameters of a kernelized low-rank feature subspace as the representation of the sequences. We cast our formulation as an optimization problem on generalized Grassmann manifolds and then solve it efficiently using Riemannian optimization techniques. We present experiments on several action recognition datasets using diverse feature modalities and demonstrate state-of-the-art results.

This paper addresses the task of dense non-rigid structure-front-motion (NRSfM) using multiple images. State-of-the-art methods to this problem are often hurdled by scalability, expensive computations, and noisy measurements. Further, recent methods to NRSfM usually either assume a small number of sparse feature points or ignore local non-linearities of shape deformations, and thus cannot reliably model complex non-rigid deformations. To address these issues, in this paper, we propose a new approach for dense NRSfM by modeling the problem on a Grassmann manifold. Specifically, we assume the complex non-rigid deformations lie on a union of local linear subspaces both spatially and temporally. This naturally allows for a compact representation of the complex non-rigid deformation over frames. We provide experimental results on several synthetic and real benchmark datasets. The procured results clearly demonstrate that our method, apart from being scalable and more accurate than state-of-the-art methods, is also more robust to noise and generalizes to highly nonlinear deformations.

Popular deep models for action recognition in videos generate independent predictions for short clips, which are then pooled heuristically to assign an action label to the full video segment. As not all frames may characterize the underlying action-indeed, many are common across multiple actions-pooling schemes that impose equal importance on all frames might be unfavorable. In an attempt to tackle this problem, we propose discriminative pooling, based on the notion that among the deep features generated on all short clips, there is at least one that characterizes the action. To this end, we learn a (nonlinear) hyperplane that separates this unknown, yet discriminative, feature from the rest. Applying multiple instance learning in a large-margin setup, we use the parameters of this separating hyperplane as a descriptor for the full video segment. Since these parameters are directly related to the support vectors in a max-margin framework, they serve as robust representations for pooling of the features. We formulate a joint objective and an efficient solver that learns these hyperplanes per video and the corresponding action classifiers over the hyperplanes. Our pooling scheme is end-to-end trainable within a deep framework. We report results from experiments on three benchmark datasets spanning a variety of challenges and demonstrate state-of-the-art performance across these tasks.

The world is fundamentally compositional, so it is natural to think of visual recognition as the recognition of basic visually primitives that are composed according to well-defined rules. This strategy allows us to recognize unseen complex concepts from simple visual primitives. However, the current trend in visual recognition follows a data greedy approach where huge amounts of data are required to learn models for any desired visual concept. In this paper, we build on the compositionality principle and develop an "algebra" to compose classifiers for complex visual concepts. To this end, we learn neural network modules to perform boolean algebra operations on simple visual classifiers. Since these modules form a complete functional set, a classifier for any complex visual concept defined as a boolean expression of primitives can be obtained by recursively applying the learned modules, even if we do not have a single training sample. As our experiments show, using such a framework, we can compose classifiers for complex visual concepts outperforming standard baselines on two well-known visual recognition benchmarks. Finally, we present a qualitative analysis of our method and its properties.

Most popular deep models for action recognition split video sequences into short sub-sequences consisting of a few frames, frame-based features are then pooled for recognizing the activity. Usually, this pooling step discards the temporal order of the frames, which could otherwise be used for better recognition. Towards this end, we propose a novel pooling method, generalized rank pooling (GRP), that takes as input, features from the intermediate layers of a CNN that is trained on tiny sub-sequences, and produces as output the parameters of a subspace which (i) provides a low-rank approximation to the features and (ii) preserves their temporal order. We propose to use these parameters as a compact representation for the video sequence, which is then used in a classification setup. We formulate an objective for computing this subspace as a Riemannian optimization problem on the Grassmann manifold, and propose an efficient conjugate gradient scheme for solving it. Experiments on several activity recognition datasets show that our scheme leads to state-of-the-art performance.

We present a principled approach to uncover the structure of visual data by solving a novel deep learning task coined visual permutation learning. The goal of this task is to find the permutation that recovers the structure of data from shuffled versions of it. In the case of natural images, this task boils down to recovering the original image from patches shuffled by an unknown permutation matrix. Unfortunately, permutation matrices are discrete, thereby posing difficulties for gradient-based methods. To this end, we resort to a continuous approximation of these matrices using doubly-stochastic matrices which we generate from standard CNN predictions using Sinkhorn iterations. Unrolling these iterations in a Sinkhorn network layer, we propose DeepPermNet, an end-to-end CNN model for this task. The utility of DeepPermNet is demonstrated on two challenging computer vision problems, namely, (i) relative attributes learning and (ii) self-supervised representation learning. Our results show state-of-the-art performance on the Public Figures and OSR benchmarks for (i) and on the classification and segmentation tasks on the PASCAL VOC dataset for (ii).

Most of the existing works on fine-grained bird image categorization and retrieval focus on finding similar images from the same species and often give little importance to inter-species similarity. In this paper, we devise a new fine-grained retrieval task that searches similar instances from different species. To this end, we propose a two-step strategy. In the first step, we search for visually similar parts to a query image using a deep convolutional neural network (CNN). To improve the quality of the retrieved candidates, we incorporate structural cues into the CNN using a novel part-pooling layer. In the second step, we re-rank the retrieved candidates improving the species diversity. We achieve this by formulating a novel ranking function that balances between the similarity of the candidates to the queried parts, while decreasing the similarity to the query species. We provide experiments on the benchmark CUB200 dataset and demonstrate clear benefits of our schemes.

Deep neural networks typically require large amounts of annotated data to be trained effectively. However, in several scientific disciplines, including medical image analysis, generating such large annotated datasets requires specialized domain knowledge, and hence is usually very expensive. In this work, we present a novel application of active learning to data sample selection for training Convolutional Neural Networks (CNN) for Cancerous Tissue Recognition (CTR). Our main idea is to steer annotation efforts towards selecting the most informative samples for training the CNN. To quantify informativeness, we explore three choices based on discrete entropy, best-vs-second-best, and k-nearest neighbor agreement. Our results on three different types of cancer datasets consistently demonstrate that under limited annotated samples, our proposed training scheme converges faster than classical randomized stochastic gradient descent, while achieving the same (or sometimes superior) classification accuracy.

Most successful deep learning algorithms for action recognition extend models designed for image-based tasks such as object recognition to video. Such extensions are typically trained for actions on single video frames or very short clips, and then their predictions from sliding-windows over the video sequence are pooled for recognizing the action at the sequence level. Usually this pooling step uses the first-order statistics of frame-level action predictions. In this paper, we explore the advantages of using higherorder correlations, specifically, we introduce Higher-order Kernel (HOK) descriptors generated from the late fusion of CNN classifier scores from all the frames in a sequence. To generate these descriptors, we use the idea of kernel linearization. Specifically, a similarity kernel matrix, which captures the temporal evolution of deep classifier scores, is first linearized into kernel feature maps. The HOK descriptors are then generated from the higher-order cooccurrences of these feature maps, and are then used as input to a video-level classifier. We provide experiments on two fine-grained action recognition datasets, and show that our scheme leads to state-of-the-art results.

Training of Convolutional Neural Networks (CNNs) on long video sequences is computationally expensive due to the substantial memory requirements and the massive number of parameters that deep architectures demand. Early fusion of video frames is thus a standard technique, in which several consecutive frames are first agglomerated into a compact representation, and then fed into the CNN as an input sample. For this purpose, a summarization approach that represents a set of consecutive RGB frames by a single dynamic image to capture pixel dynamics is proposed recently. In this paper, we introduce a novel ordered representation of consecutive optical flow frames as an alternative and argue that this representation captures the action dynamics more efficiently than RGB frames. We provide intuitions on why such a representation is better for action recognition. We validate our claims on standard benchmark datasets and demonstrate that using summaries of flow images lead to significant improvements over RGB frames while achieving accuracy comparable to the stateof-the-art on UCF101 and HMDB datasets.

Unsupervised classification plays an important role in understanding polarimetric synthetic aperture radar (PolSAR) images. One of the typical representations of PolSAR data is in the form of Hermitian positive definite (HPD) covariance matrices. Most algorithms for unsupervised classification using this representation either use statistical distribution models or adopt polarimetric target decompositions. In this paper, we propose an unsupervised classification method by introducing a sparsity-based similarity measure on HPD matrices. Specifically, we first use a novel Riemannian sparse coding scheme for representing each HPD covariance matrix as sparse linear combinations of other HPD matrices, where the sparse reconstruction loss is defined by the Riemannian geodesic distance between HPD matrices. The coefficient vectors generated by this step reflect the neighborhood structure of HPD matrices embedded in the Euclidean space and hence can be used to define a similarity measure. We apply the scheme for PolSAR data, in which we first oversegment the images into superpixels, followed by representing each superpixel by an HPD matrix. These HPD matrices are then sparse coded, and the resulting sparse coefficient vectors are then clustered by spectral clustering using the neighborhood matrix generated by our similarity measure. The experimental results on different fully PolSAR images demonstrate the superior performance of the proposed classification approach against the state-of-the-art approaches.

Data encoded as symmetric positive definite (SPD) matrices frequently arise in many areas of computer vision and machine learning. While these matrices form an open subset of the Euclidean space of symmetric matrices, viewing them through the lens of non-Euclidean Riemannian (Riem) geometry often turns out to be better suited in capturing several desirable data properties. Inspired by the great success of dictionary learning and sparse coding (DLSC) for vector-valued data, our goal in this paper is to represent data in the form of SPD matrices as sparse conic combinations of SPD atoms from a learned dictionary via a Riem geometric approach. To that end, we formulate a novel Riem optimization objective for DLSC, in which the representation loss is characterized via the affine-invariant Riem metric. We also present a computationally simple algorithm for optimizing our model. Experiments on several computer vision data sets demonstrate superior classification and retrieval performance using our approach when compared with SC via alternative non-Riem formulations.

Super-symmetric tensors - a higher-order extension of scatter matrices - are becoming increasingly popular in machine learning and computer vision for modeling data statistics, co-occurrences, or even as visual descriptors. They were shown recently to outperform second-order approaches, however, the size of these tensors are exponential in the data dimensionality, which is a significant concern. In this paper, we study third-order supersymmetric tensor descriptors in the context of dictionary learning and sparse coding. For this purpose, we propose a novel non-linear third-order texture descriptor. Our goal is to approximate these tensors as sparse conic combinations of atoms from a learned dictionary. Apart from the significant benefits to tensor compression that this framework offers, our experiments demonstrate that the sparse coefficients produced by this scheme lead to better aggregation of high-dimensional data and showcase superior performance on two common computer vision tasks compared to the state of the art.

Computer-Aided Diagnosis (CAD) has witnessed a rapid growth over the past decade, providing a variety of automated tools for the analysis of medical images. In surgical pathology, such tools enhance the diagnosing capabilities of pathologists by allowing them to review and diagnose a larger number of cases daily. Geared towards developing such tools, the main goal of this paper is to identify useful computer vision based feature descriptors for recognizing cancerous tissues in histopathologic images. To this end, we use images of Hematoxylin & Eosin-stained microscopic sections of breast and prostate carcinomas, and myometrial leiomyosarcomas, and provide an exhaustive evaluation of several state of the art feature representations for this task. Among the various image descriptors that we chose to compare, including representations based on convolutional neural networks, Fisher vectors, and sparse codes, we found that working with covariance based descriptors shows superior performance on all three types of cancer considered. While covariance descriptors are known to be effective for texture recognition, it is the first time that they are demonstrated to be useful for the proposed task and evaluated against deep learning models. Capitalizing on Region Covariance Descriptors (RCDs), we derive a powerful image descriptor for cancerous tissue recognition termed, Covariance Kernel Descriptor (CKD), which consistently outperformed all the considered image representations. Our experiments show that using CKD lead to 92.83%, 91.51%, and 98.10% classification accuracy for the recognition of breast carcinomas, prostate carcinomas, and myometrial leiomyosarcomas, respectively.

Hermitian positive definite (HPD) covariance matrices form one of the most widely-used data representations in PolSAR applications. However, most of these applications either use statistical distribution models on the PolSAR covariance matrices or polarimetric target decomposition. In this paper, we study HPD matrices for PolSAR image classification in the context of sparse coding. More specifically, the PolSAR HPD matrices are first represented as sparse linear combinations of elements from a dictionary, where each element itself is an HPD matrix and the representation loss is measured by the affine-invariant Riemannian metric. We then introduce a sparsity induced similarity measure between two HPD matrices. Finally, we propose a supervised classification scheme using support vector machines on the Riemannian sparse codes and an unsupervised classification scheme encompassing a sparsity induced similarity measure followed by spectral clustering. The proposed methods are validated on the NASA/JPL AIRSAR fully PolSAR data. The experimental results demonstrate the effectiveness of our methods.

Active Constraint Learning (ACL) is continuously gaining popularity in the area of constrained clustering due to its ability to achieve performance gains via incorporating minimal feedback from a human annotator for selected instances. For constrained clustering algorithms, such instances are integrated in the form of Must-Link (ML) and Cannot-Link (CL) constraints. Existing iterative uncertainty reduction schemes, introduce high computational burden particularly when they process larger datasets that are usually present in computer vision and visual learning applications. For scenarios that multiple agents (i.e., robots) require user feedback for performing recognition tasks, minimizing the interaction between the user and the agents, without compromising performance, is an essential task. In this study, a non-iterative ACL scheme with proven performance benefits is presented. We select to demonstrate the effectiveness of our methodology by building on the well known K-Means algorithm for clustering; one can easily extend it to alternative clustering schemes. The proposed methodology introduces the use of the Silhouette values, conventionally used for measuring clustering performance, in order to rank the degree of information content of the various samples. In addition, an efficient greedy selection scheme was devised for selecting the most informative samples for human annotation. To the best of our knowledge, this is the first active constrained clustering methodology with the ability to process computer vision datasets that this study targets. Performance results are shown on various computer vision benchmarks and support the merits of adopting the proposed scheme.

Symmetric Positive Definite (SPD) matrices emerge as data descriptors in several applications of computer vision such as object tracking, texture recognition, and diffusion tensor imaging. Clustering these data matrices forms an integral part of these applications, for which soft-clustering algorithms (K-Means, expectation maximization, etc.) are generally used. As is well-known, these algorithms need the number of clusters to be specified, which is difficult when the dataset scales. To address this issue, we resort to the classical nonparametric Bayesian framework by modeling the data as a mixture model using the Dirichlet process (DP) prior. Since these matrices do not conform to the Euclidean geometry, rather belongs to a curved Riemannian manifold,existing DP models cannot be directly applied. Thus, in this paper, we propose a novel DP mixture model framework for SPD matrices. Using the log-determinant divergence as the underlying dissimilarity measure to compare these matrices, and further using the connection between this measure and the Wishart distribution, we derive a novel DPM model based on the Wishart-Inverse-Wishart conjugate pair. We apply this model to several applications in computer vision. Our experiments demonstrate that our model is scalable to the dataset size and at the same time achieves superior accuracy compared to several state-of-the-art parametric and nonparametric clustering algorithms.

In this paper, we present a method for estimating articulated human poses in videos. We cast this as an optimization problem defined on body parts with spatio-temporal links between them. The resulting formulation is unfortunately intractable and previous approaches only provide approximate solutions. Although such methods perform well on certain body parts, e.g., head, their performance on lower arms, i.e., elbows and wrists, remains poor. We present a new approximate scheme with two steps dedicated to pose estimation. First, our approach takes into account temporal links with subsequent frames for the less-certain parts, namely elbows and wrists. Second, our method decomposes poses into limbs, generates limb sequences across time, and recomposes poses by mixing these body part sequences. We introduce a new dataset "Poses in the Wild", which is more challenging than the existing ones, with sequences containing background clutter, occlusions, and severe camera motion. We experimentally compare our method with recent approaches on this new dataset as well as on two other benchmark datasets, and show significant improvement.

In this paper, we propose a novel scheme for approximate nearest neighbor (ANN) retrieval based on dictionary learning and sparse coding. Our key innovation is to build compact codes, dubbed SpANN codes, using the active set of sparse coded data. These codes are then used to index an inverted file table for fast retrieval. The active sets are often found to be sensitive to small differences among data points, resulting in only near duplicate retrieval. We show that this sensitivity is related to the coherence of the dictionary; small coherence resulting in better retrieval. To this end, we propose a novel dictionary learning formulation with incoherence constraints and an efficient method to solve it. Experiments are conducted on two state-of-the-art computer vision datasets with 1M data points and show an order of magnitude improvement in retrieval accuracy without sacrificing memory and query time compared to the state-of-the-art methods.

This paper presents a new nearest neighbor (NN) retrieval framework: robust sparse hashing (RSH). Our approach is inspired by the success of dictionary learning for sparse coding. Our key idea is to sparse code the data using a learned dictionary, and then to generate hash codes out of these sparse codes for accurate and fast NN retrieval. But, direct application of sparse coding to NN retrieval poses a technical difficulty: when data are noisy or uncertain (which is the case with most real-world data sets), for a query point, an exact match of the hash code generated from the sparse code seldom happens, thereby breaking the NN retrieval. Borrowing ideas from robust optimization theory, we circumvent this difficulty via our novel robust dictionary learning and sparse coding framework called RSH, by learning dictionaries on the robustified counterparts of the perturbed data points. The algorithm is applied to NN retrieval on both simulated and real-world data. Our results demonstrate that RSH holds significant promise for efficient NN retrieval against the state of the art.

Covariance matrices have found success in several computer vision applications, including activity recognition, visual surveillance, and diffusion tensor imaging. This is because they provide an easy platform for fusing multiple features compactly. An important task in all of these applications is to compare two covariance matrices using a (dis)similarity function, for which the common choice is the Riemannian metric on the manifold inhabited by these matrices. As this Riemannian manifold is not flat, the dissimilarities should take into account the curvature of the manifold. As a result, such distance computations tend to slow down, especially when the matrix dimensions are large or gradients are required. Further, suitability of the metric to enable efficient nearest neighbor retrieval is an important requirement in the contemporary times of big data analytics. To alleviate these difficulties, this paper proposes a novel dissimilarity measure for covariances, the Jensen-Bregman LogDet Divergence (JBLD). This divergence enjoys several desirable theoretical properties and at the same time is computationally less demanding (compared to standard measures). Utilizing the fact that the square root of JBLD is a metric, we address the problem of efficient nearest neighbor retrieval on large covariance datasets via a metric tree data structure. To this end, we propose a K-Means clustering algorithm on JBLD. We demonstrate the superior performance of JBLD on covariance datasets from several computer vision applications.

We study Nearest Neighbors (NN) retrieval by introducing a new approach: Robust Sparse Hashing (RSH). Our approach is inspired by the success of dictionary learning for sparse coding; the key innovation is to use learned sparse codes as hashcodes for speeding up NN. But sparse coding suffers from a major drawback: when data are noisy or uncertain, for a query point, an exact match of the hashcode seldom happens, breaking the NN retrieval. We tackle this difficulty via our novel dictionary learning and sparse coding framework called RSH by learning dictionaries on the robustified counterparts of uncertain data points. The algorithm is applied to NN retrieval for Scale Invariant Feature Transform (SIFT) descriptors. The results demonstrate that RSH is noise tolerant, and at the same time shows promising NN performance over the state-of-the-art.

One of the most important tasks for mobile robots is to sense their environment. Further tasks might include the recognition of objects in the surrounding environment. Three dimensional range finders have become the sensors of choice for mapping the environment of a robot. Recognizing objects in point clouds provided by such sensors is a difficult task. The main contribution of this paper is the introduction of a new covariance based point cloud descriptor for such object recognition. Covariance based descriptors have been very successful in image processing. One of the main advantages of these descriptors is their relatively small size. The comparisons between different covariance matrices can also be made very efficient. Experiments with real world and synthetic data will show the superior performance of the covariance descriptors on point clouds compared to state-of-the-art methods.

Clinical studies confirm that mental illnesses such as autism, Obsessive Compulsive Disorder (OCD), etc. show behavioral abnormalities even at very young ages; the early diagnosis of which can help steer effective treatments. Most often, the behavior of such at-risk children deviate in very subtle ways from that of a normal child; correct diagnosis of which requires prolonged and continuous monitoring of their activities by a clinician, which is a difficult and time intensive task. As a result, the development of automation tools for assisting in such monitoring activities will be an important step towards effective utilization of the diagnostic resources. In this paper, we approach the problem from a computer vision standpoint, and propose a novel system for the automatic monitoring of the behavior of children in their natural environment through the deployment of multiple non-invasive sensors (cameras and depth sensors). We provide details of our system, together with algorithms for the robust tracking of the activities of the children. Our experiments, conducted in the Shirley G. Moore Laboratory School, demonstrate the effectiveness of our methodology.

Covariance matrices of multivariate data capture feature correlations compactly, and being very robust to noise, they have been used extensively as feature descriptors in many areas in computer vision, like, people appearance tracking, DTI imaging, face recognition, etc. Since these matrices do not adhere to the Euclidean geometry, clustering algorithms using the traditional distance measures cannot be directly extended to them. Prior work in this area has been restricted to using K-means type clustering over the Rieman-nian space using the Riemannian metric. As the applications scale, it is not practical to assume the number of components in a clustering model, failing any soft-clustering algorithm. In this paper, a novel application of the Dirich-let Process Mixture Model framework is proposed towards unsupervised clustering of symmetric positive definite matrices. We approach the problem by extending the existing K-means type clustering algorithms based on the logdet divergence measure and derive the counterpart of it in a Bayesian framework, which leads to the Wishart-Inverse Wishart conjugate pair. Alternative possibilities based on the matrix Frobenius norm and log-Euclidean measures are also proposed. The models are extensively compared using two real-world datasets against the state-of-the-art algorithms and demonstrate superior performance.

The idea of learning overcomplete dictionaries based on the paradigm of compressive sensing has found numerous applications, among which image denoising is considered one of the most successful. But many state-of-the-art denoising techniques inherently assume that the signal noise is Gaussian. We instead propose to learn overcomplete dictionaries where the signal is allowed to have both Gaussian and (sparse) Laplacian noise. Dictionary learning in this setting leads to a difficult non-convex optimization problem, which is further exacerbated by large input datasets. We tackle these difficulties by developing an efficient online algorithm that scales to data size. To assess the efficacy of our model, we apply it to dictionary learning for data that naturally satisfy our noise model, namely, Scale Invariant Feature Transform (SIFT) descriptors. For these data, we measure performance of the learned dictionary on the task of nearest-neighbor retrieval: compared to methods that do not explicitly model sparse noise our method exhibits superior performance.

Covariance matrices provide compact, informative feature descriptors for use in several computer vision applications, such as people-appearance tracking, diffusion-tensor imaging, activity recognition, among others. A key task in many of these applications is to compare different covariance matrices using a (dis)similarity function. A natural choice here is the Riemannian metric corresponding to the manifold inhabited by covariance matrices. But computations involving this metric are expensive, especially for large matrices and even more so, in gradient-based algorithms. To alleviate these difficulties, we advocate a novel dissimilarity measure for covariance matrices: the Jensen-Bregman LogDet Divergence. This divergence enjoys several useful theoretical properties, but its greatest benefits are: (i) lower computational costs (compared to standard approaches); and (ii) amenability for use in nearest-neighbor retrieval. We show numerous experiments to substantiate these claims.

Accurate localization of landmarks in the vicinity of a robot is a first step towards solving the SLAM problem. In this work, we propose algorithms to accurately estimate the 3D location of the landmarks from the robot only from a single image taken from its on board camera. Our approach differs from previous efforts in this domain in that it first reconstructs accurately the 3D environment from a single image, then it defines a coordinate system over the environment, and later it performs the desired localization with respect to this coordinate system using the environment's features. The ground plane from the given image is accurately estimated and this precedes segmentation of the image into ground and vertical regions. A Markov Random Field (MRF) based 3D reconstruction is performed to build an approximate depth map of the given image. This map is robust against texture variations due to shadows, terrain differences, etc. A texture segmentation algorithm is also applied to determine the ground plane accurately. Once the ground plane is estimated, we use the respective camera's intrinsic and extrinsic calibration information to calculate accurate 3D information about the features in the scene.

Autonomous estimation of the altitude of an Unmanned Aerial Vehicle (UAV) is extremely important when dealing with flight maneuvers like landing, steady flight, etc. Vision based techniques for solving this problem have been underutilized. In this paper, we propose a new algorithm to estimate the altitude of a UAV from top-down aerial images taken from a single on-board camera. We use a semi-supervised machine learning approach to solve the problem. The basic idea of our technique is to learn the mapping between the texture information contained in an image to a possible altitude value. We learn an over complete sparse basis set from a corpus of unlabeled images capturing the texture variations. This is followed by regression of this basis set against a training set of altitudes. Finally, a spatio-temporal Markov Random Field is modeled over the altitudes in test images, which is maximized over the posterior distribution using the MAP estimate by solving a quadratic optimization problem with L1 regularity constraints. The method is evaluated in a laboratory setting with a real helicopter and is found to provide promising results with sufficiently fast turnaround time.