Computer vision experts from around the world are gathering this week in Long Beach, California, for the 2019 Conference on Computer Vision and Pattern Recognition (CVPR). Research from Facebook AI will be presented in oral spotlight presentations and group poster sessions.
Our researchers and engineers will also be giving tutorials and participating in workshops throughout the week, including the DeepVision Workshop, co-organized by Research Scientist Cristian Canton and Chief AI Scientist Yann LeCun. We are also co-organizing the first Computer Vision for Global Challenges (CV4GC) workshop to explore new ways computer vision can be used to address global issues such as disease prevention, infrastructure mapping, and more.
Facebook AI kicked off CVPR with the Facebook AI Video Summit, organized by Jitendra Malik, Matt Feizsli, Georgia Gkioxari, Manohar Paluri, Lorenzo Torresani, and Christoph Feichtenhofer. The summit is two-day side event for a select group of researchers to have in-depth discussions on emerging technologies at the frontier of video understanding. Talks were presented by several leaders in CV from industry and academia.
For those attending CVPR, visit the Facebook Research booth to learn what we’re working on, chat about career opportunities, or try some of our demos. See the full schedule of Facebook papers, workshops, and tutorials at CVPR here.
Ruohan Gao and Kristen Grauman
Binaural audio provides a listener with 3D sound sensation, allowing a rich perceptual experience of the scene. However, binaural recordings are scarcely available and require nontrivial expertise and equipment to obtain. We propose to convert common monaural audio into binaural audio by leveraging video. The key idea is that visual frames reveal significant spatial cues that, while explicitly lacking in the accompanying single-channel audio, are strongly linked to it. Our multimodal approach recovers this link from unlabeled video. We devise a deep convolutional neural network that learns to decode the monaural (single-channel) soundtrack into its binaural counterpart by injecting visual information about object and scene configurations. We call the resulting output 2.5D visual sound — the visual stream helps “lift” the flat single channel audio into spatialized sound. In addition to sound generation, we show the self-supervised representation learned by our network benefits audio-visual source separation. Our video results: http://vision.cs.utexas.edu/projects/2.5D_visual_sound/
Read this paper’s Research in Brief here.
In this work, we demonstrate that 3D poses in video can be effectively estimated with a fully convolutional model based on dilated temporal convolutions over 2D keypoints. We also introduce back-projection, a simple and effective semisupervised training method that leverages unlabeled video data. We start with predicted 2D keypoints for unlabeled video, then estimate 3D poses, and finally back-project to the input 2D keypoints. In the supervised setting, our fully convolutional model outperforms the previous best result from the literature by 6 mm mean per-joint position error on Human3.6M, corresponding to an error reduction of 11 percent, and the model also shows significant improvements on HumanEva-I. Moreover, experiments with back-projection show that it comfortably outperforms previous state-of-the-art results in semisupervised settings where labeled data is scarce. Code and models are available at https://github.com/facebookresearch/VideoPose3D.
Zhenheng Yang, Dhruv Mahajan, Deepti Ghadiyaram, Ram Nevatia, and Vignesh Ramanathan
Weakly supervised object detection aims at reducing the amount of supervision required to train detection models. Such models are traditionally learned from images/videos labeled only with the object class and not the object bounding box. In our work, we try to leverage not only the object class labels but also the action labels associated with the data. We show that the action depicted in the image/video can provide strong cues about the location of the associated object. We learn a spatial prior for the object dependent on the action (e.g., “ball” is closer to “leg of the person” in “kicking ball”) and incorporate this prior to simultaneously train a joint object detection and action classification model. We conducted experiments on both video data sets and image data sets to evaluate the performance of our weakly supervised object detection model. Our approach outperformed the current state-of-the-art (SOTA) method by more than 6% in mAP on the Charades video data set.
Jae Sung Park, Marcus Rohrbach, Trevor Darrell, and Anna Rohrbach
While significant progress has been made in the image captioning task, video description is still in its infancy due to the complex nature of video data. Generating multi-sentence descriptions for long videos is even more challenging. Among the main issues are the fluency and coherence of the generated descriptions, and their relevance to the video. Recently, reinforcement and adversarial learning based methods have been explored to improve the image captioning models; however, both types of methods suffer from a number of issues, e.g., poor readability and high redundancy for RL and stability issues for GANs. In this work, we instead propose to apply adversarial techniques during inference, designing a discriminator which encourages better multi-sentence video description. In addition, we find that a multi-discriminator “hybrid” design, where each discriminator targets one aspect of a description, leads to the best results. Specifically, we decouple the discriminator to evaluate on three criteria: (1) visual relevance to the video; (2) language diversity and fluency; and (3) coherence across sentences. Our approach results in more accurate, diverse, and coherent multi-sentence video descriptions, as shown by automatic as well as human evaluation on the popular ActivityNet Captions data set.
Kevis-Kokitsi Maninis, Ilija Radosavovic, and Iasonas Kokkinos
In this work we address task interference in universal networks by considering that a network is trained on multiple tasks but performs one task at a time, an approach we refer to as “single-tasking multiple tasks.” The network thus modifies its behavior through task-dependent feature adaptation, or task attention. This gives the network the ability to accentuate the features that are adapted to a task while shunning irrelevant ones. We further reduce task interference by forcing the task gradients to be statistically indistinguishable through adversarial training, ensuring that the common backbone architecture serving all tasks is not dominated by any of the task-specific gradients.
Results in three multitask dense labeling problems consistently show: (1) a large reduction in the number of parameters while preserving, or even improving performance; and (2) a smooth trade-off between computation and multitask accuracy. We provide our system’s code and pretrained models at https://github.com/facebookresearch/astmt.
This paper proposes an efficient neural network (NN) architecture design methodology called Chameleon that honors given resource constraints. Instead of developing new building blocks or using computationally intensive reinforcement learning algorithms, our approach leverages existing efficient network building blocks and focuses on exploiting hardware traits and adapting computation resources to fit target latency and/or energy constraints. We formulate platform-aware NN architecture search in an optimization framework and propose a novel algorithm to search for optimal architectures aided by efficient accuracy and resource (latency and/or energy) predictors. At the core of our algorithm lies an accuracy predictor built atop Gaussian Process with Bayesian optimization for iterative sampling. With a one-time building cost for the predictors, our algorithm produces state-of-the-art model architectures on different platforms under given constraints in just minutes. Our results show that adapting computation resources to building blocks is important to model performance. Without the addition of any special features, our models achieve significant accuracy improvements relative to state-of-the-art handcrafted and automatically designed architectures. We achieve 73.8% and 75.3% top-1 accuracy on ImageNet at 20ms latency on a mobile CPU and DSP. At reduced latency, our models achieve up to 8.2% (4.8%) and 6.7% (9.3%) absolute top-1 accuracy improvements compared to MobileNetV2 and MnasNet, respectively, on a mobile CPU (DSP), and 2.7% (4.6%) and 5.6% (2.6%) accuracy gains over ResNet-101 and ResNet-152, respectively, on an Nvidia GPU (Intel CPU).
Despite significant progress in Visual Question Answering over the years, robustness of today’s VQA models leave much to be desired. We introduce a new evaluation protocol and associated data set (VQA-Rephrasings) and show that state-of-the-art VQA models are notoriously brittle to linguistic variations in questions. VQA-Rephrasings contains three human-provided rephrasings for 40K questions spanning 40K images from the VQA v2.0 validation data set. As a step towards improving robustness of VQA models, we propose a model-agnostic framework that exploits cycle consistency. Specifically, we train a model to not only answer a question but also generate a question conditioned on the answer, such that the answer predicted for the generated question is the same as the ground truth answer to the original question. Without the use of additional annotations, we show that our approach is significantly more robust to linguistic variations than state-of-the-art VQA models, when evaluated on the VQA-Rephrasings data set. In addition, our approach outperforms state-of-the-art approaches on the standard VQA and Visual Question Generation tasks on the challenging VQA v2.0 data set.
Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove
Computer graphics, 3D computer vision, and robotics communities have produced multiple approaches to representing 3D geometry for rendering and reconstruction. These provide trade-offs across fidelity, efficiency, and compression capabilities. In this work, we introduce DeepSDF, a learned continuous Signed Distance Function (SDF) representation of a class of shapes that enables high-quality shape representation, interpolation, and completion from partial and noisy 3D input data. DeepSDF, like its classical counterpart, represents a shape’s surface by a continuous volumetric field: The magnitude of a point in the field represents the distance to the surface boundary, and the sign indicates whether the region is inside (-) or outside (+) of the shape, hence our representation implicitly encodes a shape’s boundary as the zero-level-set of the learned function while explicitly representing the classification of space as being part of the shape’s interior or not. While classical SDFs both in analytical or discretized voxel form typically represent the surface of a single shape, DeepSDF can represent an entire class of shapes. Furthermore, we show state-of-the-art performance for learned 3D shape representation and completion while reducing the model size by an order of magnitude compared with previous work.
A plethora of recent work has shown that convolutional networks are not robust to adversarial images: images that are created by perturbing a sample from the data distribution as to maximize the loss on the perturbed example. In this work, we hypothesize that adversarial perturbations move the image away from the image manifold in the sense that there exists no physical process that could have produced the adversarial image. This hypothesis suggests that a successful defense mechanism against adversarial images should aim to project the images back onto the image manifold. We study such defense mechanisms, which approximate the projection onto the unknown image manifold by a nearest-neighbor search against a web-scale image database containing tens of billions of images. Empirical evaluations of this defense strategy on ImageNet suggest that it is very effective in attack settings in which the adversary does not have access to the image database. We also propose two novel attack methods to break nearest-neighbor defenses, and demonstrate conditions under which nearest-neighbor defense fails. We perform a series of ablation experiments, which suggest that there is a trade-off between robustness and accuracy in our defenses, that a large image database (with hundreds of millions of images) is crucial to get good performance, and that careful construction the image database is important to be robust against attacks tailored to circumvent our defenses.
Motion has shown to be useful for video understanding, where motion is typically represented by optical flow. However, computing flow from video frames is very time-consuming. Recent works directly leverage the motion vectors and residuals readily available in the compressed video to represent motion at no performance cost. While this avoids flow computation, it also hurts accuracy since the motion vector is noisy and has substantially reduced resolution, which makes it a less discriminative motion representation. To remedy these issues, we propose a lightweight generator network, which reduces noises in motion vectors and captures fine motion details, achieving a more Discriminative Motion Cue (DMC) representation. Since optical flow is a more accurate motion representation, we train the DMC generator to approximate flow using a reconstruction loss and an adversarial loss, jointly with the downstream action classification task. Extensive evaluations on three action recognition benchmarks (HMDB-51, UCF-101, and a subset of Kinetics) confirm the effectiveness of our method. Our full system, consisting of the generator and the classifier, is coined as DMC-Net which obtains high accuracy close to that of using flow and runs two orders of magnitude faster than using optical flow at inference time.
To help bridge the gap between internet vision-style problems and the goal of vision for embodied perception we instantiate a large-scale navigation task – Embodied Question Answering  in photo-realistic environments (Matterport3D). We thoroughly study navigation policies that utilize 3D point clouds, RGB images, or their combination. Our analysis of these models reveals several key findings. We find that two seemingly naive navigation baselines, forward-only and random, are strong navigators and challenging to outperform, due to the specific choice of the evaluation setting presented by . We find a novel loss-weighting scheme we call Inflection Weighting to be important when training recurrent models for navigation with behavior cloning and are able to out perform the baselines with this technique. We find that point clouds provide a richer signal than RGB images for learning obstacle avoidance, motivating the use (and continued study) of 3D deep learning models for embodied navigation.
Standard image captioning tasks such as COCO and Flickr30k are factual, neutral in tone and (to a human) state the obvious (e.g., “a man playing a guitar”). While such tasks are useful to verify that a machine understands the content of an image, they are not engaging to humans as captions. With this in mind we define a new task, PERSONALITY-CAPTIONS, where the goal is to be as engaging to humans as possible by incorporating controllable style and personality traits. We collect and release a large data set of 241,858 of such captions conditioned over 215 possible traits. We build models that combine existing work from (1) sentence representations  with Transformers trained on 1.7 billion dialogue examples; and (2) image representations  with ResNets trained on 3.5 billion social media images. We obtain state-of-the-art performance on Flickr30k and COCO, and strong performance on our new task. Finally, online evaluations validate that our task and models are engaging to humans, with our best model close to human performance.
Zhenpei Yang, Jeffrey Z. Pan, Linjie Luo, Xiaowei Zhou, Kristen Grauman, and Qixing Huang
Estimating the relative rigid pose between two RGB-D scans of the same underlying environment is a fundamental problem in computer vision, robotics, and computer graphics. Most existing approaches allow only limited relative pose changes since they require considerable overlap between the input scans. We introduce a novel approach that extends the scope to extreme relative poses, with little or even no overlap between the input scans. The key idea is to infer more complete scene information about the underlying environment and match on the completed scans. In particular, instead of only performing scene completion from each individual scan, our approach alternates between relative pose estimation and scene completion. This allows us to perform scene completion by utilizing information from both input scans at late iterations, resulting in better results for both scene completion and relative pose estimation. Experimental results on benchmark data sets show that our approach leads to considerable improvements over state-of-the-art approaches for relative pose estimation. In particular, our approach provides encouraging relative pose estimates even between non-overlapping scans.
Designing accurate and efficient ConvNets for mobile devices is challenging because the design space is combinatorially large. Due to this, previous neural architecture search (NAS) methods are computationally expensive. ConvNet architecture optimality depends on factors such as input resolution and target devices. However, existing approaches are too resource demanding for case-by-case redesigns. Also, previous work focuses primarily on reducing FLOPs, but FLOP count does not always reflect actual latency. To address these, we propose a differentiable neural architecture search (DNAS) framework that uses gradient-based methods to optimize ConvNet architectures, avoiding enumerating and training individual architectures separately as in previous methods. FBNets (Facebook-Berkeley-Nets), a family of models discovered by DNAS surpass state-of-the-art models both designed manually and generated automatically. FBNet-B achieves 74.1% top-1 accuracy on ImageNet with 295M FLOPs and 23.1 ms latency on a Samsung S8 phone, 2.4x smaller and 1.5x faster than MobileNetV2-1.3 with similar accuracy. Despite higher accuracy and lower latency than MnasNet, we estimate FBNet-B’s search cost is 420x smaller than MnasNet’s, at only 216 GPUhours. Searched for different resolutions and channel sizes, FBNets achieve 1.5% to 6.4% higher accuracy than MobileNetV2. The smallest FBNet achieves 50.2% accuracy and 2.9 ms latency (345 frames per second) on a Samsung S8. Over a Samsung-optimized FBNet, the iPhone-X-optimized model achieves a 1.4x speedup on an iPhone X. FBNet models are open-sourced at https://github.com/facebookresearch/mobile-vision.
Read this paper's Research in Brief here.
Adversarial attacks to image classification systems present challenges to convolutional networks and opportunities for understanding them. This study suggests that adversarial perturbations on images lead to noise in the features constructed by these networks. Motivated by this observation, we develop new network architectures that increase adversarial robustness by performing feature denoising. Specifically, our networks contain blocks that denoise the features using non-local means or other filters; the entire networks are trained end-to-end. When combined with adversarial training, our feature denoising networks substantially improve the state of the art in adversarial robustness in both white-box and black-box attack settings. On ImageNet, under 10-iteration PGD white-box attacks where prior art has 27.9% accuracy, our method achieves 55.7%; even under extreme 2000-iteration PGD white-box attacks, our method secures 42.6% accuracy. Our method was ranked first in Competition on Adversarial Attacks and Defenses (CAAD) 2018 — it achieved 50.6% classification accuracy on a secret, ImageNet-like test data set against 48 unknown attackers, surpassing the runner-up approach by approximately 10%. Code is available at https://github.com/facebookresearch/ImageNet-Adversarial-Training.
Read this paper's Research in Brief here.
Globally modeling and reasoning over relations between regions can be beneficial for many computer vision tasks on both images and videos. Convolutional neural networks (CNNs) excel at modeling local relations by convolution operations, but they are typically inefficient at capturing global relations between distant regions and require stacking multiple convolution layers. In this work, we propose a new approach for reasoning globally in which a set of features are globally aggregated over the coordinate space and then projected to an interaction space where relational reasoning can be efficiently computed. After reasoning, relation-aware features are distributed back to the original coordinate space for down-stream tasks. We further present a highly efficient instantiation of the proposed approach and introduce the Global Reasoning unit (GloRe unit) that implements the coordinate-interaction space mapping by weighted global pooling and weighted broadcasting, and the relation reasoning via graph convolution on a small graph in interaction space. The proposed GloRe unit is lightweight, end-to-end trainable and can be easily plugged into existing CNNs for a wide range of tasks. Extensive experiments show our GloRe unit can consistently boost the performance of state-of-the-art backbone architectures, including ResNet [15, 16], ResNeXt , SE-Net  and DPN , for both 2D and 3D CNNs, on image classification, semantic segmentation and video action recognition task.
Video description is one of the most challenging problems in vision and language understanding due to the large variability both on the video and language side. Models, hence, typically shortcut the difficulty in recognition and generate plausible sentences that are based on priors but are not necessarily grounded in the video. In this work, we explicitly link the sentence to the evidence in the video by annotating each noun phrase in a sentence with the corresponding bounding box in one of the frames of a video. Our data set, ActivityNet-Entities, augments the challenging ActivityNet Captions data set with 158K bounding box annotations, each grounding a noun phrase. This allows training video description models with this data, and importantly, evaluate how grounded or “true” such model are to the video they describe. To generate grounded captions, we propose a novel video description model which is able to exploit these bounding box annotations. We demonstrate the effectiveness of our model on our data set, but also show how it can be applied to image description on the Flickr30k Entities data set. We achieve state-of-the-art performance on video description, video paragraph description, and image description and demonstrate our generated sentences are better grounded in the video.
Road network extraction from satellite images often produce fragmented road segments leading to road maps unfit for real applications. Pixel-wise classification fails to predict topologically correct and connected road masks due to the absence of connectivity supervision and difficulty in enforcing topological constraints. In this paper, we propose a connectivity task called Orientation Learning, motivated by the human behavior of annotating roads by tracing it at a specific orientation. We also develop a stacked multi-branch convolutional module to effectively utilize the mutual information between orientation learning and segmentation tasks. These contributions ensure that the model predicts topologically correct and connected road masks. We also propose Connectivity Refinement approach to further enhance the estimated road networks. The refinement model is pretrained to connect and refine the corrupted ground-truth masks and later fine-tuned to enhance the predicted road masks. We demonstrate the advantages of our approach on two diverse road extraction data sets SpaceNet  and DeepGlobe . Our approach improves over the state-of-the-art techniques by 9% and 7.5% in road topology metric on SpaceNet and DeepGlobe, respectively.
Amaia Salvador, Michal Drozdzal, Xavier Giro-i-Nieto, and Adriana Romero
People enjoy food photography because they appreciate food. Behind each meal there is a story described in a complex recipe and, unfortunately, by simply looking at a food image we do not have access to its preparation process. Therefore, in this paper we introduce an inverse cooking system that recreates cooking recipes given food images. Our system predicts ingredients as sets by means of a novel architecture, modeling their dependencies without imposing any order, and then generates cooking instructions by attending to both image and its inferred ingredients simultaneously. We extensively evaluate the whole system on the large-scale Recipe1M data set and show that (1) we improve performance w.r.t. previous baselines for ingredient prediction; (2) we are able to obtain high-quality recipes by leveraging both image and ingredients; and (3) our system is able to produce more compelling recipes than retrieval-based approaches according to human judgment. We make code and models publicly available at: https://github.com/facebookresearch/inversecooking.
Read this paper's Research in Brief here
Dejan Azinovic, Tzu-Mao Li, Anton Kaplanyan, and Matthias Niessner
Modern computer vision algorithms have brought significant advancement to 3D geometry reconstruction. However, illumination and material reconstruction remain less studied, with current approaches assuming very simplified models for materials and illumination. We introduce Inverse Path Tracing, a novel approach to jointly estimate the material properties of objects and light sources in indoor scenes by using an invertible light transport simulation. We assume a coarse geometry scan, along with corresponding images and camera poses. The key contribution of this work is an accurate and simultaneous retrieval of light sources and physically based material properties (e.g., diffuse reflectance, specular reflectance, roughness, etc.) for the purpose of editing and re-rendering the scene under new conditions. To this end, we introduce a novel optimization method using a differentiable Monte Carlo renderer that computes derivatives with respect to the estimated unknown illumination and material properties. This enables joint optimization for physically correct light transport and material models using a tailored stochastic gradient descent.
Yu-Chuan Su and Kristen Grauman
Ideally, 360◦ imagery could inherit the deep convolutional neural networks (CNNs) already trained with great success on perspective projection images. However, existing methods to transfer CNNs from perspective to spherical images introduce significant computational costs and/or degradations in accuracy. We present the Kernel Transformer Network (KTN) to efficiently transfer convolution kernels from perspective images to the equirectangular projection of 360◦ images. Given a source CNN for perspective images as input, the KTN produces a function parameterized by a polar angle and kernel as output. Given a novel 360◦ image, that function in turn can compute convolutions for arbitrary layers and kernels as would the source CNN on the corresponding tangent plane projections. Distinct from all existing methods, KTNs allow model transfer: The same model can be applied to different source CNNs with the same base architecture. This enables application to multiple recognition tasks without retraining the KTN. Validating our approach with multiple source CNNs and data sets, we show that KTNs improve the state of the art for spherical convolution. KTNs successfully preserve the source CNN’s accuracy while offering transferability, scalability to typical image resolutions, and, in many cases, a substantially lower memory footprint.
Current fully supervised video data sets consist of only a few hundred thousand videos and fewer than a thousand domain-specific labels. This hinders the progress towards advanced video architectures. This paper presents an in-depth study of using large volumes of web videos for pretraining video models for the task of action recognition. Our primary empirical finding is that pretraining at a very large scale (over 65 million videos), despite on noisy social-media videos and hashtags, substantially improves the state of the art on three challenging public action recognition data sets. Further, we examine three questions in the construction of weakly supervised video action data sets. First, given that actions involve interactions with objects, how should one construct a verb-object pretraining label space to benefit transfer learning the most? Second, frame-based models perform quite well on action recognition; is pretraining for good image features sufficient or is pretraining for spatio-temporal features valuable for optimal transfer learning? Finally, actions are generally less well localized in long videos vs. short videos; since action labels are provided at a video level, how should one choose video clips for best performance, given some fixed budget of number or minutes of videos?
Chun-Liang Li, Tomas Simon, Jason Saragih, Barnabás Póczos, and Yaser Sheikh
We present LBS-AE, a self-supervised autoencoding algorithm for fitting articulated mesh models to point clouds. As input, we take a sequence of point clouds to be registered as well as an artist-rigged mesh, i.e., a template mesh equipped with a linear-blend skinning (LBS) deformation space parameterized by a skeleton hierarchy. As output, we learn an LBS-based autoencoder that produces registered meshes from the input point clouds. To bridge the gap between the artist-defined geometry and the captured point clouds, our autoencoder models pose-dependent deviations from the template geometry. During training, instead of using explicit correspondences, such as key points or pose supervision, our method leverages LBS deformations to bootstrap the learning process. To avoid poor local minima from erroneous point-to-point correspondences, we utilize a structured Chamfer distance based on part-segmentations, which are learned concurrently using self-supervision. We demonstrate qualitative results on real captured hands and report quantitative evaluations on the FAUST benchmark for body registration. Our method achieves performance that is superior to other unsupervised approaches and comparable to methods using supervised examples.
Bo Xiong, Yannis Kalantidis, Deepti Ghadiyaram, and Kristen Grauman
Highlight detection has the potential to significantly ease video browsing, but existing methods often suffer from expensive supervision requirements, where human viewers must manually identify highlights in training videos. We propose a scalable unsupervised solution that exploits video duration as an implicit supervision signal. Our key insight is that video segments from shorter user-generated videos are more likely to be highlights than those from longer videos, since users tend to be more selective about the content when capturing shorter videos. Leveraging this insight, we introduce a novel ranking framework that prefers segments from shorter videos while properly accounting for the inherent noise in the (unlabeled) training data. We use it to train a highlight detector with 10M hashtagged Instagram videos. In experiments on two challenging public video highlight detection benchmarks, our method substantially improves the state of the art for unsupervised highlight detection.
To understand the world, we humans constantly need to relate the present to the past, and put events in context. In this paper, we enable existing video models to do the same. We propose a long-term feature bank — supportive information extracted over the entire span of a video — to augment state-of-the-art video models that otherwise would only view short clips of 2-5 seconds. Our experiments demonstrate that augmenting 3D convolutional networks with a long-term feature bank yields state-of-the-art results on three challenging video data sets: AVA, EPIC-Kitchens, and Charades. Code is available online.
Progress on object detection is enabled by data sets that focus the research community’s attention on open challenges. This process led us from simple images to complex scenes and from bounding boxes to segmentation masks. In this work, we introduce LVIS (pronounced “el-vis”): a new data set for Large Vocabulary Instance Segmentation. We plan to collect 2.2 million high-quality instance segmentation masks for over 1,000 entry-level object categories in 164K images. Due to the Zipfian distribution of categories in natural images, LVIS naturally has a long tail of categories with few training samples. Given that state-of-the-art deep learning methods for object detection perform poorly in the low-sample regime, we believe that our data set poses an important and exciting new scientific challenge. LVIS is available at http://www.lvisdata set.org.
Embodied Question Answering (EQA) is a relatively new task where an agent is asked to answer questions about its environment from egocentric perception. EQA as introduced in  makes the fundamental assumption that every question, e.g. “what color is the car?”, has exactly one target (“car”) being inquired about. This assumption puts a direct limitation on the abilities of the agent.
We present a generalization of EQA – Multi-Target EQA (MT-EQA). Specifically, we study questions that have multiple targets in them, such as “Is the dresser in the bedroom bigger than the oven in the kitchen?”, where the agent has to navigate to multiple locations (“dresser in bedroom”, “oven in kitchen”) and perform comparative reasoning (“dresser” bigger than “oven”) before it can answer a question. Such questions require the development of entirely new modules or components in the agent. To address this, we propose a modular architecture composed of a program generator, a controller, a navigator, and a VQA module. The program generator converts the given question into sequential executable sub-programs; the navigator guides the agent to multiple locations pertinent to the navigation-related sub-programs; and the controller learns to select relevant observations along its path. These observations are then fed to the VQA module to predict the answer. We perform detailed analysis for each of the model components and show that our joint model can outperform previous methods and strong baselines by a significant margin. Project page: https://embodiedqa.org.
Yedid Hoshen and Jitendra Malik
Unconditional image generation has recently been dominated by generative adversarial networks (GANs). GAN methods train a generator which regresses images from random noise vectors, as well as a discriminator that attempts to differentiate between the generated images and a training set of real images. GANs have shown amazing results at generating realistic looking images. Despite their success, GANs suffer from critical drawbacks, including unstable training and mode-dropping. The weaknesses in GANs have motivated research into alternatives including: variational auto-encoders (VAEs), latent embedding learning methods (e.g., GLO), and nearest-neighbor based implicit maximum likelihood estimation (IMLE). Unfortunately at the moment, GANs still significantly outperform the alternative methods for image generation. In this work, we present a novel method — Generative Latent Nearest Neighbors (GLANN) — for training generative models without adversarial training. GLANN combines the strengths of IMLE and GLO in a way that overcomes the main drawbacks of each method. Consequently, GLANN generates images that are far better than GLO and IMLE. Our method does not suffer from mode collapse which plagues GAN training and is much more stable. Qualitative results show that GLANN outperforms a baseline consisting of 800 GANs and VAEs on commonly used data sets. Our models are also shown to be effective for training truly non-adversarial unsupervised image translation.
The recently introduced panoptic segmentation task has renewed our community’s interest in unifying the tasks of instance segmentation (for thing classes) and semantic segmentation (for stuff classes). However, current state-of-the-art methods for this joint task use separate and dissimilar networks for instance and semantic segmentation, without performing any shared computation. In this work, we aim to unify these methods at the architectural level, designing a single network for both tasks. Our approach is to endow Mask R-CNN, a popular instance segmentation method, with a semantic segmentation branch using a shared Feature Pyramid Network (FPN) backbone. Surprisingly, this simple baseline not only remains effective for instance segmentation but also yields a lightweight, top-performing method for semantic segmentation. In this work, we perform a detailed study of this minimally extended version of Mask R-CNN with FPN, which we refer to as Panoptic FPN, and show it is a robust and accurate baseline for both tasks. Given its effectiveness and conceptual simplicity, we hope our method can serve as a strong baseline and aid future research in panoptic segmentation.
We propose and study a task we name panoptic segmentation (PS). Panoptic segmentation unifies the typically distinct tasks of semantic segmentation (assign a class label to each pixel) and instance segmentation (detect and segment each object instance). The proposed task requires generating a coherent scene segmentation that is rich and complete, an important step toward real-world vision systems. While early work in computer vision addressed related image/scene parsing tasks, these are not currently popular, possibly due to lack of appropriate metrics or associated recognition challenges. To address this, we propose a novel panoptic quality (PQ) metric that captures performance for all classes (stuff and things) in an interpretable and unified manner. Using the proposed metric, we perform a rigorous study of both human and machine performance for PS on three existing data sets, revealing interesting insights about the task. The aim of our work is to revive the interest of the community in a more unified view of image segmentation.
The goal of MRI reconstruction is to restore a high-fidelity image from partially observed measurements. This partial view naturally induces reconstruction uncertainty that can only be reduced by acquiring additional measurements. In this paper, we present a novel method for MRI reconstruction that, at inference time, dynamically selects the measurements to take and iteratively refines the prediction in order to best reduce the reconstruction error and, thus, its uncertainty. We validate our method on a large-scale knee MRI data set as well as on ImageNet. Results show that (1) our system successfully outperforms active acquisition baselines; (2) our uncertainty estimates correlate with error maps; and (3) our ResNet-based architecture surpasses standard pixel-to-pixel models in the task of MRI reconstruction. The proposed method not only shows high-quality reconstructions but also paves the road toward more applicable solutions for accelerating MRI.
Read this paper's Research in Brief here.
Jae Shin Yoon, Takaaki Shiratori, Shoou-I Yu, and Hyun Soo Park
Improvements in data capture and face modeling techniques have enabled us to create high-fidelity realistic face models. However, driving these realistic face models requires special input data, e.g., 3D meshes and unwrapped textures. Also, these face models expect clean input data taken under controlled lab environments, which is very different from data collected in the wild. All these constraints make it challenging to use the high-fidelity models in tracking for commodity cameras. In this paper, we propose a self-supervised domain adaptation approach to enable the animation of high-fidelity face models from a commodity camera. Our approach first circumvents the requirement for special input data by training a new network that can directly drive a face model just from a single 2D image. Then, we overcome the domain mismatch between lab and uncontrolled environments by performing self-supervised domain adaptation based on “consecutive frame texture consistency” based on the assumption that the appearance of the face is consistent over consecutive frames, avoiding the necessity of modeling the new environment such as lighting or background. Experiments show that we are able to drive a high-fidelity face model to perform complex facial motion from a cellphone camera without requiring any labeled data from the new domain.
DensePose supersedes traditional landmark detectors by densely mapping image pixels to body surface coordinates. This power, however, comes at a greatly increased annotation cost, as supervising the model requires to manually label hundreds of points per pose instance. In this work, we thus seek methods to significantly slim down the DensePose annotations, proposing more efficient data collection strategies. In particular, we demonstrate that if annotations are collected in video frames, their efficacy can be multiplied for free by using motion cues. To explore this idea, we introduce DensePose-Track, a data set of videos where selected frames are annotated in the traditional DensePose manner. Then, building on geometric properties of the DensePose mapping, we use the video dynamic to propagate ground-truth annotations in time as well as to learn from Siamese equivariance constraints. Having performed exhaustive empirical evaluation of various data annotation and learning strategies, we demonstrate that doing so can deliver significantly improved pose estimation results over strong baselines. However, despite what is suggested by some recent works, we show that merely synthesizing motion patterns by applying geometric transformations to isolated frames is significantly less effective, and that motion cues help much more when they are extracted from videos.
Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, and Henry Fuchs
We propose a system that uses a convolution neural network (CNN) to estimate depth from a stereo pair followed by volumetric fusion of the predicted depth maps to produce a 3D reconstruction of a scene. Our proposed depth refinement architecture predicts view-consistent disparity and occlusion maps that help the fusion system to produce geometrically consistent reconstructions. We utilize 3D dilated convolutions in our proposed cost filtering network that yields better filtering while almost halving the computational cost in comparison to state-of-the-art cost filtering architectures. For feature extraction we use the Vortex Pooling architecture. The proposed method achieves state-of-the-art results in KITTI 2012, KITTI 2015, and ETH 3D stereo benchmarks. Finally, we demonstrate that our system is able to produce high-fidelity 3D scene reconstructions that outperform the state-of-the-art stereo system.
Giljoo Nam, Chenglei Wu, Min H. Kim, and Yaser Sheikh
Hair is one of the most challenging objects to reconstruct due to its micro-scale structure and a large number of repeated strands with heavy occlusions. In this paper, we present the first method to capture high-fidelity hair geometry with strand-level accuracy. Our method takes three stages to achieve this. In the first stage, a new multi-view stereo method with a slanted support line is proposed to solve the hair correspondences between different views. In detail, we contribute a novel cost function consisting of both photo-consistency term and geometric term that reconstructs each hair pixel as a 3D line. By merging all the depth maps, a point cloud, as well as local line directions for each point, is obtained. Thus, in the second stage, we feature a novel strand reconstruction method with the mean-shift to convert the noisy point data to a set of strands. Lastly, we grow the hair strands with multi-view geometric constraints to elongate the short strands and recover the missing strands, thus significantly increasing the reconstruction completeness. We evaluate our method on both synthetic data and real captured data, showing that our method can reconstruct hair strands with sub-millimeter accuracy.
Aron Yu and Kristen Grauman
Current wisdom suggests more labeled image data is always better, and obtaining labels is the bottleneck. Yet curating a pool of sufficiently diverse and informative images is itself a challenge. In particular, training image curation is problematic for fine-grained attributes, where the subtle visual differences of interest may be rare within traditional image sources. We propose an active image generation approach to address this issue. The main idea is to jointly learn the attribute ranking task while also learning to generate novel realistic image samples that will benefit that task. We introduce an end-to-end framework that dynamically “imagines” image pairs that would confuse the current model, presents them to human annotators for labeling, then improves the predictive model with the new examples. On two data sets, we show that by thinking outside the pool of real images, our approach gains generalization accuracy on challenging fine-grained attribute comparisons.
Studies have shown that a dominant class of questions asked by visually impaired users on images of their surroundings involves reading text in the image. But today’s VQA models can not read! Our paper takes a first step towards addressing this problem. First, we introduce a new “TextVQA” data set to facilitate progress on this important problem. Existing data sets either have a small proportion of questions about text (e.g., the VQA data set) or are too small (e.g., the VizWiz data set). TextVQA contains 45,336 questions on 28,408 images that require reasoning about text to answer. Second, we introduce a novel model architecture that reads text in the image, reasons about it in the context of the image and the question, and predicts an answer which might be a deduction based on the text and the image or is composed of the strings found in the image. Consequently, we call our approach Look, Read, Reason & Answer (LoRRA). We show that LoRRA outperforms existing state-of-the-art VQA models on our TextVQA data set. We find that the gap between human performance and machine performance is significantly larger on TextVQA than on VQA 2.0, suggesting that TextVQA is well suited to benchmark progress along directions complementary to VQA 2.0.
See codebase on GitHub here.
For full lists of organizers and speakers, see the CVPR website.
Bringing Robots to the Computer Vision Community (Tutorial)
Abhinav Gupta, invited speaker
Saurabh Gupta, organizer
Computer Vision After 5 Years Workshop
Saurabh Gupta, organizer
Practical 3D Photography (demo)
Computer Vision for Global Challenges (CV4GC) Workshop
Laura Sevilla-Lara and Yannis Kalantidis, lead organizers
Egocentric Perception, Interaction, and Computing (EPIC) Workshop
Kristen Grauman, organizer
Habitat: Embodied Agents Challenge and Workshop
Richard Newcombe and Jitendra Malik, speakers
Manolis Savva, Abhishek Kadian, Oleksandr Maksymets, Julian Straub, Bhavana Jain, Devi Parikh, Georgia Gkioxari, Marcus Rohrbach, Jitendra Malik, Erik Wijmans, and Dhruv Batra, organizers
Workshop on Media Forensics
Cristian Canton, organizer
Precognition: Seeing through the Future Workshop
Paper: Leveraging the Present to Anticipate the Future in Videos
Antoine Miech, Ivan Laptev, Josef Sivic, Heng Wang, Lorenzo Torresani, and Du Tran
Sight and Sound Workshop
Kristen Grauman, organizer
Understanding Subjective Attributes of Data: Focus on Fashion and Subjective Search
Kristen Grauman and Devi Parikh, organizers