ICCV 2019 Papers

We study the video super-resolution (SR) problem for facilitating video analytics tasks, e.g. action recognition, instead of for visual quality. The popular action recognition methods based on convolutional networks, exemplified by two-stream networks, are not directly applicable on video of low spatial resolution. This can be remedied by performing video SR prior to recognition, which motivates us to improve the SR procedure for recognition accuracy. Tailored for two-stream action recognition networks, we propose two video SR methods for the spatial and temporal streams respectively. On the one hand, we observe that regions with action are more important to recognition, and we propose an optical-flow guided weighted mean-squared-error loss for our spatial-oriented SR (SoSR) network to emphasize the reconstruction of moving objects. On the other hand, we observe that existing video SR methods incur temporal discontinuity between frames, which also worsens the recognition accuracy, and we propose a siamese network for our temporal-oriented SR (ToSR) training that emphasizes the temporal continuity between consecutive frames. We perform experiments using two state-of-the-art action recognition networks and two well-known datasets--UCF101 and HMDB51. Results demonstrate the effectiveness of our proposed SoSR and ToSR in improving recognition accuracy.

We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D object detection model that learns a 3D object detector network from training data with two disjoint sets of object classes - a set of strong classes with both 2D and 3D box labels, and another set of weak classes with only 2D box labels. In particular, we suggest a relaxed reprojection loss, box prior loss and a Box-to-Point Cloud Fit network that allow us to effectively transfer useful 3D information from the strong classes to the weak classes during training, and consequently, enable the network to detect 3D objects in the weak classes during inference. Experimental results show that our proposed algorithm outperforms baseline approaches and achieves promising results compared to fully-supervised approaches on the SUN-RGBD and KITTI datasets. Furthermore, we show that our Box-to-Point Cloud Fit network improves performances of the fully-supervised approaches on both datasets.

Vision-and-Language Navigation (VLN) tasks such as Room-to-Room (R2R) require machine agents to interpret natural language instructions and learn to act in visually realistic environments to achieve navigation goals. The overall task requires competence in several perception problems: successful agents combine spatio-temporal, vision and language understanding to produce appropriate action sequences. Our approach adapts pre-trained vision and language representations to relevant in-domain tasks making them more effective for VLN. Specifically, the representations are adapted to solve both a cross-modal sequence alignment and sequence coherence task. In the sequence alignment task, the model determines whether an instruction corresponds to a sequence of visual frames. In the sequence coherence task, the model determines whether the perceptual sequences are predictive sequentially in the instruction-conditioned latent space. By transferring the domain-adapted representations, we improve competitive agents in R2R as measured by the success rate weighted by path length (SPL) metric.

Zero-shot learning (ZSL) is a challenging problem that aims to recognize the target categories without seen data, where semantic information is leveraged to transfer knowledge from some source classes. Although ZSL has made great progress in recent years, most existing approaches are easy to overfit the sources classes in generalized zero-shot learning (GZSL) task, which indicates that they learn little knowledge about target classes. To tackle such problem, we propose a novel Transferable Contrastive Network (TCN) that explicitly transfers knowledge from the source classes to the target classes. It automatically contrasts one image with different classes to judge whether they are consistent or not. By exploiting the class similarities to make knowledge transfer from source images to similar target classes, our approach is more robust to recognize the target images. Experiments on five benchmark datasets show the superiority of our approach for GZSL.

Few-shot learning, which aims at extracting new concepts rapidly from extremely few examples of novel classes, has been featured into the meta-learning paradigm recently. Yet, the key challenge of how to learn a generalizable classifier with the capability of adapting to specific tasks with severely limited data still remains in this domain. To this end, we propose a Transductive Episodic-wise Adaptive Metric (TEAM) framework for few-shot learning, by integrating the meta-learning paradigm with both deep metric learning and transductive inference. With exploring the pairwise constraints and regularization prior within each task, we explicitly formulate the adaptation procedure into a standard semi-definite programming problem. By solving the problem with its closed-form solution on the fly with the setup of transduction, our approach efficiently tailors an episodic-wise metric for each task to adapt all features from a shared task-agnostic embedding space into a more discriminative task-specific metric space. Moreover, we further leverage an attention-based bi-directional similarity strategy for extracting the more robust relationship between queries and prototypes. Extensive experiments on three benchmark datasets show that our framework is superior to other existing approaches and achieves the state-of-the-art performance in the few-shot literature.

The problem of tracking multiple objects in a video sequence poses several challenging tasks. For tracking-by-detection, these include object re-identification, motion prediction and dealing with occlusions. We present a tracker (without bells and whistles) that accomplishes tracking without specifically targeting any of these tasks, in particular, we perform no training or optimization on tracking data. To this end, we exploit the bounding box regression of an object detector to predict the position of an object in the next frame, thereby converting a detector into a Tracktor. We demonstrate the potential of Tracktor and provide a new state-of-the-art on three multi-object tracking benchmarks by extending it with a straightforward re-identification and camera motion compensation. We then perform an analysis on the performance and failure cases of several state-of-the-art tracking methods in comparison to our Tracktor. Surprisingly, none of the dedicated tracking methods are considerably better in dealing with complex tracking scenarios, namely, small and occluded objects or missing detections. However, our approach tackles most of the easy tracking scenarios. Therefore, we motivate our approach as a new tracking paradigm and point out promising future research directions. Overall, Tracktor yields superior tracking performance than any current tracking method and our analysis exposes remaining and unsolved tracking challenges to inspire future research directions.

The performance of deep neural networks is strongly influenced by the quantity and quality of annotated data. Most of the large activity recognition datasets consist of data sourced from the web, which does not reflect challenges that exist in activities of daily living. In this paper, we introduce a large real-world video dataset for activities of daily living: Toyota Smarthome. The dataset consists of 16K RGB+D clips of 31 activity classes, performed by seniors in a smarthome. Unlike previous datasets, videos were fully unscripted. As a result, the dataset poses several challenges: high intra-class variation, high class imbalance, simple and composite activities, and activities with similar motion and variable duration. Activities were annotated with both coarse and fine-grained labels. These characteristics differentiate Toyota Smarthome from other datasets for activity recognition. As recent activity recognition approaches fail to address the challenges posed by Toyota Smarthome, we present a novel activity recognition method with attention mechanism. We propose a pose driven spatio-temporal attention mechanism through 3D ConvNets. We show that our novel method outperforms state-of-the-art methods on benchmark datasets, as well as on the Toyota Smarthome dataset. We release the dataset for research use.

Understanding images without explicit supervision has become an important problem in computer vision. In this paper, we address image captioning by generating language descriptions of scenes without learning from annotated pairs of images and their captions. The core component of our approach is a shared latent space that is structured by visual concepts. In this space, the two modalities should be indistinguishable. A language model is first trained to encode sentences into semantically structured embeddings. Image features that are translated into this embedding space can be decoded into descriptions through the same language model, similarly to sentence embeddings. This translation is learned from weakly paired images and text using a loss robust to noisy assignments and a conditional adversarial component. Our approach allows to exploit large text corpora outside the annotated distributions of image/caption data. Our experiments show that the proposed domain alignment learns a semantically meaningful representation which outperforms previous work.

It is challenging for weakly supervised object detection network to precisely predict the positions of the objects, since there are no instance-level category annotations. Most existing methods tend to solve this problem by using a two-phase learning procedure, i.e., multiple instance learning detector followed by a fully supervised learning detector with bounding-box regression. Based on our observation, this procedure may lead to local minima for some object categories. In this paper, we propose to jointly train the two phases in an end-to-end manner to tackle this problem. Specifically, we design a single network with both multiple instance learning and bounding-box regression branches that share the same backbone. Meanwhile, a guided attention module using classification loss is added to the backbone for effectively extracting the implicit location information in the features. Experimental results on public datasets show that our method achieves state-of-the-art performance.

Virtual try-on systems under arbitrary human poses have significant application potential, yet also raise extensive challenges, such as self-occlusions, heavy misalignment among different poses, and complex clothes textures. Existing virtual try-on methods can only transfer clothes given a fixed human pose, and still show unsatisfactory performances, often failing to preserve person identity or texture details, and with limited pose diversity. This paper makes the first attempt towards a multi-pose guided virtual try-on system, which enables clothes to transfer onto a person with diverse poses. Given an input person image, a desired clothes image, and a desired pose, the proposed Multi-pose Guided Virtual Try-On Network (MG-VTON) generates a new person image after fitting the desired clothes into the person and manipulating the pose. MG-VTON is constructed with three stages: 1) a conditional human parsing network is proposed that matches both the desired pose and the desired clothes shape; 2) a deep Warping Generative Adversarial Network (Warp-GAN) that warps the desired clothes appearance into the synthesized human parsing map and alleviates the misalignment problem between the input human pose and the desired one; 3) a refinement render network recovers the texture details of clothes and removes artifacts, based on multi-pose composition masks. Extensive experiments on commonly-used datasets and our newly-collected largest virtual try-on benchmark demonstrate that our MG-VTON significantly outperforms all state-of-the-art methods both qualitatively and quantitatively, showing promising virtual try-on performances.

This paper addresses the task of learning latent attributes from triplet similarity comparisons. Consider, for instance, the three shoes in Fig. 1(a). They can be compared according to color, comfort, size, or shape resulting in different rankings. Most approaches for embedding learning either make a simplifying assumption - that all inputs are comparable under a single criterion, or require expensive attribute supervision. We introduce Latent Similarity Networks (LSNs): a simple and effective technique to discover the underlying latent notions of similarity in data without any explicit attribute supervision. LSNs can be trained with standard triplet supervision and learn several latent embeddings that can be used to compare images under multiple notions of similarity. LSNs achieve state-of-the-art performance on UT-Zappos-50k Shoes and Celeb-A Faces datasets and also demonstrate the ability to uncover meaningful latent attributes.

Deep neural network based methods have made a significant breakthrough in salient object detection. However, they are typically limited to input images with low resolutions (400x400 pixels or less). Little effort has been made to train neural networks to directly handle salient object segmentation in high-resolution images. This paper pushes forward high-resolution saliency detection, and contributes a new dataset, named High-Resolution Salient Object Detection (HRSOD) dataset. To our best knowledge, HRSOD is the first high-resolution saliency detection dataset to date. As another contribution, we also propose a novel approach, which incorporates both global semantic information and local high-resolution details, to address this challenging task. More specifically, our approach consists of a Global Semantic Network (GSN), a Local Refinement Network (LRN) and a Global-Local Fusion Network (GLFN). The GSN extracts the global semantic information based on downsampled entire image. Guided by the results of GSN, the LRN focuses on some local regions and progressively produces high-resolution predictions. The GLFN is further proposed to enforce spatial consistency and boost performance. Experiments illustrate that our method outperforms existing state-of-the-art methods on high-resolution saliency datasets by a large margin, and achieves comparable or even better performance than them on some widely used saliency benchmarks.

Aggregating multi-level features is essential for capturing multi-scale context information for precise scene semantic segmentation. However, the improvement by directly fusing shallow features and deep features becomes limited as the semantic gap between them increases. To solve this problem, we explore two strategies for robust feature fusion. One is enhancing shallow features using a semantic enhancement module (SeEM) to alleviate the semantic gap between shallow features and deep features. The other strategy is feature attention, which involves discovering complementary information (i.e., boundary information) from low-level features to enhance high-level features for precise segmentation. By embedding these two strategies, we construct a parallel feature pyramid towards improving multi-level feature fusion. A Semantic Enhanced Network called SeENet is constructed with the parallel pyramid to implement precise segmentation. Experiments on three benchmark datasets demonstrate the effectiveness of our method for robust multi-level feature aggregation. As a result, our SeENet has achieved better performance than other state-of-the-art methods for semantic segmentation.

Object detection is an important vision task and has emerged as an indispensable component in many vision system, rendering its robustness as an increasingly important performance factor for practical applications. While object detection models have been demonstrated to be vulnerable against adversarial attacks by many recent works, very few efforts have been devoted to improving their robustness. In this work, we take an initial attempt towards this direction. We first revisit and systematically analyze object detectors and many recently developed attacks from the perspective of model robustness. We then present a multi-task learning perspective of object detection and identify an asymmetric role of task losses. We further develop an adversarial training approach which can leverage the multiple sources of attacks for improving the robustness of detection models. Extensive experiments on PASCAL-VOC and MS-COCO verified the effectiveness of the proposed approach.

We show that denoising of 3D point clouds can be learned unsupervised, directly from noisy 3D point cloud data only. This is achieved by extending recent ideas from learning of unsupervised image denoisers to unstructured 3D point clouds. Unsupervised image denoisers operate under the assumption that a noisy pixel observation is a random realization of a distribution around a clean pixel value, which allows appropriate learning on this distribution to eventually converge to the correct value. Regrettably, this assumption is not valid for unstructured points: 3D point clouds are subject to total noise, i.e. deviations in all coordinates, with no reliable pixel grid. Thus, an observation can be the realization of an entire manifold of clean 3D points, which makes the quality of a naive extension of unsupervised image denoisers to 3D point clouds unfortunately only little better than mean filtering. To overcome this, and to enable effective and unsupervised 3D point cloud denoising, we introduce a spatial prior term, that steers converges to the unique closest out of the many possible modes on the manifold. Our results demonstrate unsupervised denoising performance similar to that of supervised learning with clean data when given enough training examples - whereby we do not need any pairs of noisy and clean training data.

We propose a new approach, named PolyMapper, to circumvent the conventional pixel-wise segmentation of (aerial) images and predict objects in a vector representation directly. PolyMapper directly extracts the topological map of a city from overhead images as collections of building footprints and road networks. In order to unify the shape representation for different types of objects, we also propose a novel sequentialization method that reformulates a graph structure as closed polygons. Experiments are conducted on both existing and self-collected large-scale datasets of several cities. Our empirical results demonstrate that our end-to-end learnable model is capable of drawing polygons of building footprints and road networks that very closely approximate the structure of existing online map services, in a fully automated manner. Quantitative and qualitative comparison to the state-of-the-arts also show that our approach achieves good levels of performance. To the best of our knowledge, the automatic extraction of large-scale topological maps is a novel contribution in the remote sensing community that we believe will help develop models with more informed geometrical constraints.

Real-time generic object detection on mobile platforms is a crucial but challenging computer vision task. Prior lightweight CNN-based detectors are inclined to use one-stage pipeline. In this paper, we investigate the effectiveness of two-stage detectors in real-time generic detection and propose a lightweight two-stage detector named ThunderNet. In the backbone part, we analyze the drawbacks in previous lightweight backbones and present a lightweight backbone designed for object detection. In the detection part, we exploit an extremely efficient RPN and detection head design. To generate more discriminative feature representation, we design two efficient architecture blocks, Context Enhancement Module and Spatial Attention Module. At last, we investigate the balance between the input resolution, the backbone, and the detection head. Benefit from the highly efficient backbone and detection part design, ThunderNet surpasses previous lightweight one-stage detectors with only 40% of the computational cost on PASCAL VOC and COCO benchmarks. Without bells and whistles, ThunderNet runs at 24.1 fps on an ARM-based device with 19.2 AP on COCO. To the best of our knowledge, this is the first real-time detector reported on ARM platforms. Code will be released for paper reproduction.

This paper presents RF-Avatar, a neural network model that can estimate 3D meshes of the human body in the presence of occlusions, baggy clothes, and bad lighting conditions. We leverage that radio frequency (RF) signals in the WiFi range traverse clothes and occlusions and bounce off the human body. Our model parses such radio signals and recovers 3D body meshes. Our meshes are dynamic and smoothly track the movements of the corresponding people. Further, our model works both in single and multi-person scenarios. Inferring body meshes from radio signals is a highly under-constrained problem. Our model deals with this challenge using: 1) a combination of strong and weak supervision, 2) a multi-headed self-attention mechanism that attends differently to temporal information in the radio signal, and 3) an adversarially trained temporal discriminator that imposes a prior on the dynamics of human motion. Our results show that RF-Avatar accurately recovers dynamic 3D meshes in the presence of occlusions, baggy clothes, bad lighting conditions, and even through walls.

We present the first method to perform automatic 3D pose, shape and texture capture of animals from images acquired in-the-wild. In particular, we focus on the problem of capturing 3D information about Grevy's zebras from a collection of images. The Grevy's zebra is one of the most endangered species in Africa, with only a few thousand individuals left. Capturing the shape and pose of these animals can provide biologists and conservationists with information about animal health and behavior. In contrast to research on human pose, shape and texture estimation, training data for endangered species is limited, the animals are in complex natural scenes with occlusion, they are naturally camouflaged, travel in herds, and look similar to each other. To overcome these challenges, we integrate the recent SMAL animal model into a network-based regression pipeline, which we train end-to-end on synthetically generated images with pose, shape, and background variation. Going beyond state-of-the-art methods for human shape and pose estimation, our method learns a shape space for zebras during training. Learning such a shape space from images using only a photometric loss is novel, and the approach can be used to learn shape in other settings with limited 3D supervision. Moreover, we couple 3D pose and shape prediction with the task of texture synthesis, obtaining a full texture map of the animal from a single image. We show that the predicted texture map allows a novel per-instance unsupervised optimization over the network features. This method, SMALST (SMAL with learned Shape and Texture) goes beyond previous work, which assumed manual keypoints and/or segmentation, to regress directly from pixels to 3D animal shape, pose and texture.