CVPR 2020 Papers

Effective defense of deep neural networks against adversarial attacks remains a challenging problem, especially under powerful white-box attacks. In this paper, we develop a new method called ensemble generative cleaning with feedback loops (EGC-FL) for effective defense of deep neural networks. The proposed EGC-FL method is based on two central ideas. First, we introduce a transformed deadzone layer into the defense network, which consists of an orthonormal transform and a deadzone-based activation function, to destroy the sophisticated noise pattern of adversarial attacks. Second, by constructing a generative cleaning network with a feedback loop, we are able to generate an ensemble of diverse estimations of the original clean image. We then learn a network to fuse this set of diverse estimations together to restore the original image. Our extensive experimental results demonstrate that our approach improves the state-of-art by large margins in both white-box and black-box attacks. It significantly improves the classification accuracy for white-box PGD attacks upon the second best method by more than 29% on the SVHN dataset and more than 39% on the challenging CIFAR-10 dataset.

Whereas adversarial training is employed as the main defence strategy against specific adversarial samples, it has limited generalization capability and incurs excessive time complexity. In this paper, we propose an attack-agnostic defence framework to enhance the intrinsic robustness of neural networks, without jeopardizing the ability of generalizing clean samples. Our Feature Pyramid Decoder (FPD) framework applies to all block-based convolutional neural networks (CNNs). It implants denoising and image restoration modules into a targeted CNN, and it also constraints the Lipschitz constant of the classification layer. Moreover, we propose a two-phase strategy to train the FPD-enhanced CNN, utilizing e-neighbourhood noisy images with multi-task and self-supervised learning. Evaluated against a variety of white-box and black-box attacks, we demonstrate that FPD-enhanced CNNs gain sufficient robustness against general adversarial samples on MNIST, SVHN and CALTECH. In addition, if we further conduct adversarial training, the FPD-enhanced CNNs perform better than their non-enhanced versions.

Deep learning approaches to generic (non-semantic) segmentation have so far been indirect and relied on edge detection. This is in contrast to semantic segmentation, where DNNs are applied directly. We propose an alternative approach called Deep Generic Segmentation (DGS) and try to follow the path used for semantic segmentation. Our main contribution is a new method for learning a pixel-wise representation that reflects segment relatedness. This representation is combined with a CRF to yield the segmentation algorithm. We show that we are able to learn meaningful representations that improve segmentation quality and that the representations themselves achieve state-of-the-art segment similarity scores. The segmentation results are competitive and promising.

Neural networks are known to be vulnerable to carefully crafted adversarial examples, and these malicious samples often transfer, i.e., they remain adversarial even against other models. Although significant effort has been devoted to the transferability across models, surprisingly little attention has been paid to cross-task transferability, which represents the real-world cybercriminal's situation, where an ensemble of different defense/detection mechanisms need to be evaded all at once. We investigate the transferability of adversarial examples across a wide range of real-world computer vision tasks, including image classification, object detection, semantic segmentation, explicit content detection, and text detection. Our proposed attack minimizes the "dispersion" of the internal feature map, overcoming the limitations of existing attacks, that require task-specific loss functions and/or probing a target model. We conduct evaluation on open-source detection and segmentation models, as well as four different computer vision tasks provided by Google Cloud Vision (GCV) APIs. We demonstrate that our approach outperforms existing attacks by degrading performance of multiple CV tasks by a large margin with only modest perturbations.

Unsupervised domain adaptation (UDA) is a representative problem in transfer learning, which aims to improve the classification performance on an unlabeled target domain by exploiting discriminant information from a labeled source domain. The optimal transport model has been used for UDA in the perspective of distribution matching. However, the transport distance cannot reflect the discriminant information from either domain knowledge or category prior. In this work, we propose an enhanced transport distance (ETD) for UDA. This method builds an attention-aware transport distance, which can be viewed as the prediction feedback of the iteratively learned classifier, to measure the domain discrepancy. Further, the Kantorovich potential variable is re-parameterized by deep neural networks to learn the distribution in the latent space. The entropy-based regularization is developed to explore the intrinsic structure of the target domain. The proposed method is optimized alternately in an end-to-end manner. Extensive experiments are conducted on four benchmark datasets to demonstrate the SOTA performance of ETD.

Reliable and accurate 3D object detection is a necessity for safe autonomous driving. Although LiDAR sensors can provide accurate 3D point cloud estimates of the environment, they are also prohibitively expensive for many settings. Recently, the introduction of pseudo-LiDAR (PL) has led to a drastic reduction in the accuracy gap between methods based on LiDAR sensors and those based on cheap stereo cameras. PL combines state-of-the-art deep neural networks for 3D depth estimation with those for 3D object detection by converting 2D depth map outputs to 3D point cloud inputs. However, so far these two networks have to be trained separately. In this paper, we introduce a new framework based on differentiable Change of Representation (CoR) modules that allow the entire PL pipeline to be trained end-to-end. The resulting framework is compatible with most state-of-the-art networks for both tasks and in combination with PointRCNN improves over PL consistently across all benchmarks --- yielding the highest entry on the KITTI image-based 3D object detection leaderboard at the time of submission. Our code will be made available at https://github.com/mileyan/pseudo-LiDAR_e2e.

Reinforcement Learning (RL) aims at learning an optimal behavior policy from its own experiments and not rule-based control methods. However, there is no RL algorithm yet capable of handling a task as difficult as urban driving. We present a novel technique, coined implicit affordances, to effectively leverage RL for urban driving thus including lane keeping, pedestrians and vehicles avoidance, and traffic light detection. To our knowledge we are the first to present a successful RL agent handling such a complex task especially regarding the traffic light detection. Furthermore, we have demonstrated the effectiveness of our method by winning the Camera Only track of the CARLA challenge.

In this work, we propose an end-to-end framework to learn local multi-view descriptors for 3D point clouds. To adopt a similar multi-view representation, existing studies use hand-crafted viewpoints for rendering in a preprocessing stage, which is detached from the subsequent descriptor learning stage. In our framework, we integrate the multi-view rendering into neural networks by using a differentiable renderer, which allows the viewpoints to be optimizable parameters for capturing more informative local context of interest points. To obtain discriminative descriptors, we also design a soft-view pooling module to attentively fuse convolutional features across views. Extensive experiments on existing 3D registration benchmarks show that our method outperforms existing local descriptors both quantitatively and qualitatively.

Deep networks excel in learning patterns from large amounts of data. On the other hand, many geometric vision tasks are specified as optimization problems. To seamlessly combine deep learning and geometric vision, it is vital to perform learning and geometric optimization end-to-end. Towards this aim, we present BPnP, a novel network module that backpropagates gradients through a Perspective-n-Points (PnP) solver to guide parameter updates of a neural network. Based on implicit differentiation, we show that the gradients of a "self-contained" PnP solver can be derived accurately and efficiently, as if the optimizer block were a differentiable function. We validate BPnP by incorporating it in a deep model that can learn camera intrinsics, camera extrinsics (poses) and 3D structure from training datasets. Further, we develop an end-to-end trainable pipeline for object pose estimation, which achieves greater accuracy by combining feature-based heatmap losses with 2D-3D reprojection errors. Since our approach can be extended to other optimization problems, our work helps to pave the way to perform learnable geometric vision in a principled manner. Our PyTorch implementation of BPnP is available on http://github.com/BoChenYS/BPnP.

Previous deep learning approaches to color constancy usually directly estimate illuminant value from input image. Such approaches might suffer heavily from being sensitive to the variation of image content. To overcome this problem, we introduce a deep metric learning approach named Illuminant-Guided Triplet Network (IGTN) to color constancy. IGTN generates an Illuminant Consistent and Discriminative Feature (ICDF) for achieving robust and accurate illuminant color estimation. ICDF is composed of semantic and color features based on a learnable color histogram scheme. In the ICDF space, regardless of the similarities of their contents, images taken under the same or similar illuminants are placed close to each other and at the same time images taken under different illuminants are placed far apart. We also adopt an end-to-end training strategy to simultaneously group image features and estimate illuminant value, and thus our approach does not have to classify illuminant in a separate module. We evaluate our method on two public datasets and demonstrate our method outperforms state-of-the-art approaches. Furthermore, we demonstrate that our method is less sensitive to image appearances, and can achieve more robust and consistent results than other methods on a High Dynamic Range dataset.

The increasing number of vision-based tracking systems deployed in production have necessitated fast, robust camera calibration. In the domain of sport, the majority of current work focuses on sports where lines and intersections are easy to extract, and appearance is relatively consistent across venues. However, for more challenging sports like basketball, those techniques are not sufficient. In this paper, we propose an end-to-end approach for single moving camera calibration across challenging scenarios in sports. Our method contains three key modules: 1) area-based court segmentation, 2) camera pose estimation with embedded templates, 3) homography prediction via a spatial transform network (STN). All three modules are connected, enabling end-to-end training. We evaluate our method on a new college basketball dataset and demonstrate state of the art performance in variable and dynamic environments. We also validate our method on the World Cup 2014 dataset to show its competitive performance against the state-of-the-art methods. Lastly, we show that our method is two orders of magnitude faster than the previous state of the art on both datasets.

Multi-modal clustering aims to cluster data into different groups by exploring complementary information from multiple modalities or views. Little work learns the deep fused representations and simutaneously discovers the cluster structure with a discriminative loss. In this paper, we present an End-to-end Adversarial-attention network for Multi-modal Clustering (EAMC), where adversarial learning and attention mechanism are leveraged to align the latent feature distributions and quantify the importance of modalities respectively. To benefit from the joint training, we introducea divergence-based clustering objective that not only encourages the separation and compactness of the clusters but also enjoy a clear cluster structure by embedding the simplex geometry of the output space into the loss. The proposed network consists of modality-specific feature learning, modality fusion and cluster assignment three modules. It can be trained from scratch with batch-mode based optimization and avoid an autoencoder pretraining stage. Comprehensive experiments conducted on five real-world datasets show the superiority and effectiveness of the proposed clustering method.

3D instance segmentation plays a predominant role in environment perception of robotics and augmented reality. Many deep learning based methods have been presented recently for this task. These methods rely on either a detection branch to propose objects or a grouping step to assemble same-instance points. However, detection based methods do not ensure a consistent instance label for each point, while the grouping step requires parameter-tuning and is computationally expensive. In this paper, we introduce a novel framework to enable end-to-end instance segmentation without detection and a separate step of grouping. The core idea is to convert instance segmentation to a candidate assignment problem. At first, a set of instance candidates is sampled. Then we propose an assignment module for candidate assignment and a suppression module to eliminate redundant candidates. A mapping between instance labels and instance candidates is further sought to construct an instance grouping loss for the network training. Experimental results demonstrate that our method is more effective and efficient than previous approaches.

We present EmotiCon, a learning-based algorithm for context-aware perceived human emotion recognition from videos and images. Motivated by Frege's Context Principle from psychology, our approach combines three interpretations of context for emotion recognition. Our first interpretation is based on using multiple modalities (e.g.faces and gaits) for emotion recognition. For the second interpretation, we gather semantic context from the input image and use a self-attention-based CNN to encode this information. Finally, we use depth maps to model the third interpretation related to socio-dynamic interactions and proximity among agents. We demonstrate the efficiency of our network through experiments on EMOTIC, a benchmark dataset. We report an Average Precision (AP) score of 35.48 across 26 classes, which is an improvement of 7-8 over prior methods. We also introduce a new dataset, GroupWalk, which is a collection of videos captured in multiple real-world settings of people walking. We report an AP of 65.83 across 4 categories on GroupWalk, which is also an improvement over prior methods.

We propose associating language utterances to 3D visual abstractions of the scene they describe. The 3D visual abstractions are encoded as 3-dimensional visual feature maps. We infer these 3D visual scene feature maps from RGB images of the scene via view prediction: when the generated 3D scene feature map is neurally projected from a camera viewpoint, it should match the corresponding RGB image. We present generative models that condition on the dependency tree of an utterance and generate a corresponding visual 3D feature map as well as reason about its plausibility, and detector models that condition on both the dependency tree of an utterance and a related image and localize the object referents in the 3D feature map inferred from the image. Our model outperforms models of language and vision that associate language with 2D CNN activations or 2D images by a large margin in a variety of tasks, such as, classifying plausibility of utterances, detecting referential expressions, and supplying rewards for trajectory optimization of object placement policies from language instructions. We perform numerous ablations and show the improved performance of our detectors is due to its better generalization across camera viewpoints and lack of object interferences in the inferred 3D feature space, and the improved performance of our generators is due to their ability to spatially reason about objects and their configurations in 3D when mapping from language to scenes.

Learning the distance metric between pairs of samples has been studied for image retrieval and clustering. With the remarkable success of pair-based metric learning losses, recent works have proposed the use of generated synthetic points on metric learning losses for augmentation and generalization. However, these methods require additional generative networks along with the main network, which can lead to a larger model size, slower training speed, and harder optimization. Meanwhile, post-processing techniques, such as query expansion and database augmentation, have proposed the combination of feature points to obtain additional semantic information. In this paper, inspired by query expansion and database augmentation, we propose an augmentation method in an embedding space for pair-based metric learning losses, called embedding expansion. The proposed method generates synthetic points containing augmented information by a combination of feature points and performs hard negative pair mining to learn with the most informative feature representations. Because of its simplicity and flexibility, it can be used for existing metric learning losses without affecting model size, training speed, or optimization difficulty. Finally, the combination of embedding expansion and representative metric learning losses outperforms the state-of-the-art losses and previous sample generation methods in both image retrieval and clustering tasks. The implementation is publicly available.

Model efficiency has become increasingly important in computer vision. In this paper, we systematically study neural network architecture design choices for object detection and propose several key optimizations to improve efficiency. First, we propose a weighted bi-directional feature pyramid network (BiFPN), which allows easy and fast multi-scale feature fusion; Second, we propose a compound scaling method that uniformly scales the resolution, depth, and width for all backbone, feature network, and box/class prediction networks at the same time. Based on these optimizations and EfficientNet backbones, we have developed a new family of object detectors, called EfficientDet, which consistently achieve much better efficiency than prior art across a wide spectrum of resource constraints. In particular, with single-model and single-scale, our EfficientDetD7 achieves state-of-the-art 52.2 AP on COCO test-dev with 52M parameters and 325B FLOPs, being 4x - 9x smaller and using 13x - 42x fewer FLOPs than previous detector.

In this work, we introduce a novel local pairwise descriptor and then develop a simple, effective iterative method to solve the resulting quadratic assignment through sparsity control for shape correspondence between two approximate isometric surfaces. Our pairwise descriptor is based on the stiffness and mass matrix of finite element approximation of the Laplace-Beltrami differential operator, which is local in space, sparse to represent, and extremely easy to compute while containing global information. It allows us to deal with open surfaces, partial matching, and topological perturbations robustly. To solve the resulting quadratic assignment problem efficiently, the two key ideas of our iterative algorithm are: 1) select pairs with good (approximate) correspondence as anchor points, 2) solve a regularized quadratic assignment problem only in the neighborhood of selected anchor points through sparsity control. These two ingredients can improve and increase the number of anchor points quickly while reducing the computation cost in each quadratic assignment iteration significantly. With enough high-quality anchor points, one may use various pointwise global features with reference to these anchor points to further improve the dense shape correspondence. We use various experiments to show the efficiency, quality, and versatility of our method on large data sets, patches, and point clouds (without global meshes).

In order to remove the non-uniform blur of images captured from dynamic scenes, many deep learning based methods design deep networks for large receptive fields and strong fitting capabilities, or use multi-scale strategy to deblur image on different scales gradually. Restricted by the fixed structures and parameters, these methods are always huge in model size to handle complex blurs. In this paper, we start from the deblurring deconvolution operation, then design an effective and real-time deblurring network. The main contributions are three folded, 1) we construct a spatially variant deconvolution network using modulated deformable convolutions, which can adjust receptive fields adaptively according to the blur features. 2) our analysis shows the sampling points of deformable convolution can be used to approximate the blur kernel, which can be simplified to bi-directional optical flows. So the position learning of sampling points can be supervised by bi-directional optical flows. 3) we build a light-weighted backbone for image restoration problem, which can balance the calculations and effectiveness well. Experimental results show that the proposed method achieves state-of-the-art deblurring performance, but with less parameters and shorter running time.