ECCV 2018 论文列表

Current methods for single-image depth estimation use training datasets with real image-depth pairs or stereo pairs, which are not easy to acquire. We propose a framework, trained on synthetic image-depth pairs and unpaired real images, that comprises an image translation network for enhancing realism of input images, followed by a depth prediction network. A key idea is having the first network act as a wide-spectrum input translator, taking in either synthetic or real images, and ideally producing minimally modified realistic images. This is done via a reconstruction loss when the training input is real, and a GAN loss when synthetic, removing the need for heuristic self-regularization. The second network is trained on a task loss for synthetic image-depth pairs, with an extra GAN loss to unify real and synthetic feature distributions. Importantly, the framework can be trained end-to-end, leading to good results, even surpassing early deep-learning methods that use real paired data.

We address the problem of image feature learning for scene text recognition. The image features in the state-of-the-art methods are learned from large-scale synthetic image datasets. However, most methods only rely on outputs of the synthetic data generation process, namely realistically looking images, and completely ignore the rest of the process. We propose to leverage the parameters that lead to the output images to improve image feature learning. Specifically, for every image out of the data generation process, we obtain the associated parameters and render another "clean" image that is free of select distortion factors that are applied to the output image. Because of the absence of distortion factors, the clean image tends to be easier to recognize than the original image. We design a multi-task network with an encoder-discriminator-generator architecture to guide the feature of the original image toward that of the clean image. The experiments show that our method significantly outperforms the state-of-the-art methods on standard scene text recognition benchmarks. Furthermore, we show that without explicitly handling, our method works on challenging cases where input images contain severe geometric distortion, such as text on a curved path.

Videos contain highly redundant information between frames. Such redundancy has been studied extensively in video compression and encoding but is less explored for more advanced video processing. In this paper, we propose a learnable unified framework for propagating a variety of visual properties of video images, including but not limited to color, high dynamic range (HDR), and segmentation mask, where the properties are available for only a few key-frames. Our approach is based on a temporal propagation network (TPN), which models the transition-related affinity between a pair of frames in a purely data-driven manner. We theoretically prove two essential properties of TPN: (a) by regularizing the global transformation matrix as orthogonal, the ``style energy'' of the property can be well preserved during propagation; and (b) such regularization can be achieved by the proposed switchable TPN with bi-directional training on pairs of frames. We apply the switchable TPN to three tasks: colorizing a gray-scale video based on a few colored key-frames, generating an HDR video from a low dynamic range (LDR) video and a few HDR frames, and propagating a segmentation mask from the first frame in videos. Experimental results show that our approach is significantly more accurate and efficient than the state-of-the-art methods.

We present SwapNet, a framework to transfer garments across images of people with arbitrary body pose, shape, and clothing. Garment transfer is a challenging task that requires (i) disentangling the features of the clothing from the body pose and shape and (ii) realistic synthesis of the garment texture on the new body. We present a neural network architecture that tackles these sub-problems with two task-specific sub-networks. Since acquiring pairs of images showing the same clothing on different bodies is difficult, we propose a novel weakly-supervised approach that generates training pairs from a single image via data augmentation. We present the first fully automatic method for garment transfer in unconstrained images without solving the difficult 3D reconstruction problem. We demonstrate a variety of transfer results and highlight our advantages over traditional image-to-image and analogy pipelines.

Recent work has demonstrated that it is possible to learn deep neural networks for monocular depth and ego-motion estimation from unlabelled video sequences, an interesting theoretical development with numerous advantages in applications. In this paper, we propose a number of improvements to these approaches. First, since such self-supervised approaches are based on the brightness constancy assumption, which is valid only for a subset of pixels, we propose a probabilistic learning formulation where the network predicts distributions over variables rather than specific values. As these distributions are conditioned on the observed image, the network can learn which scene and object types are likely to violate the model assumptions, resulting in more robust learning. We also propose to build on dozens of years of experience in developing handcrafted structure-from-motion (SFM) algorithms. We do so by using an off-the-shelf SFM system to generate a supervisory signal for the deep neural network. While this signal is also noisy, we show that our probabilistic formulation can learn and account for the defects of SFM, helping to integrate different sources of information and boosting the overall performance of the network.

Superpixels provide an efficient low/mid-level representation of image data, which greatly reduces the number of image primitives for subsequent vision tasks. Existing superpixel algorithms are not differentiable, making them difficult to integrate into otherwise end-to-end trainable deep neural networks. We develop a new differentiable model for superpixel sampling that leverages deep networks for learning superpixel segmentation. The resulting "Superpixel Sampling Network" (SSN) is end-to-end trainable, which allows learning task-specific superpixels with flexible loss functions and has fast runtime. Extensive experimental analysis indicates that SSNs not only outperform existing superpixel algorithms on traditional segmentation benchmarks, but can also learn superpixels for other tasks. In addition, SSNs can be easily integrated into downstream deep networks resulting in performance improvements.

We present a flexible framework for robust computed tomography (CT) reconstruction with a specific emphasis on recovering thin 1D and 2D manifolds embedded in 3D volumes. To reconstruct such structures at resolutions below the Nyquist limit of the CT image sensor, we devise a new 3D structure tensor prior, which can be incorporated as a regularizer into more traditional proximal optimization methods for CT reconstruction. As a second, smaller contribution, we also show that when using such a proximal reconstruction framework, it is beneficial to employ the Simultaneous Algebraic Reconstruction Technique (SART) instead of the commonly used Conjugate Gradient (CG) method in the solution of the data term proximal operator. We show empirically that CG often does not converge to the global optimum for tomography problem even though the underlying problem is convex. We demonstrate that using SART provides better reconstruction results in sparse-view settings using fewer projection images. We provide extensive experimental results for both contributions on both simulated and real data. Moreover, our code will also be made publicly available.

Face hallucination is a generative task to super-resolve the facial image with low resolution while human perception of face heavily relies on identity information. However, previous face hallucination approaches largely ignore facial identity recovery. This paper proposes Super-Identity Convolutional Neural Network (SICNN) to recover identity information for generating faces closed to the real identity. Specifically, we define a super-identity loss to measure the identity difference between a hallucinated face and its corresponding high-resolution face within the hypersphere identity metric space. However, directly using this loss will lead to a Dynamic Domain Divergence problem, which is caused by the large margin between the high-resolution domain and the hallucination domain. To overcome this challenge, we present a domain-integrated training approach by constructing a robust identity metric for faces from these two domains. Extensive experimental evaluations demonstrate that the proposed SICNN achieves superior visual quality over the state-of-the-art methods on a challenging task to super-resolve 12$ imes$14 faces with an 8$ imes$ upscaling factor. In addition, SICNN significantly improves the recognizability of ultra-low-resolution faces.

Video highlight or summarization is among interesting topics in computer vision, which benefits a variety of applications like viewing, searching, or storage. However, most existing studies rely on training data of third-person videos, which cannot easily generalize to highlight the first-person ones. With the goal of deriving an effective model to summarize first-person videos, we propose a novel deep neural network architecture for describing and discriminating vital spatiotemporal information across videos with different points of view. Our proposed model is realized in a semi-supervised setting, in which fully annotated third-person videos, unlabeled first-person videos, and a small amount of annotated first-person ones are presented during training. In our experiments, qualitative and quantitative evaluations on both benchmarks and our collected first-person video datasets are presented.

The recent years have witnessed significant growth in constructing robust generative models to capture informative distributions of natural data. However, it is difficult to fully exploit the distribution of complex data, like images and videos, due to the high dimensionality of ambient space. Sequentially, how to effectively guide the training of generative models is a crucial issue. In this paper, we present a subspace-based generative adversarial network (Sub-GAN) which simultaneously disentangles multiple latent subspaces and generates diverse samples correspondingly. Since the high-dimensional natural data usually lies on a union of low-dimensional subspaces which contain semantically extensive structure, Sub-GAN incorporates a novel clusterer that can interact with the generator and discriminator via subspace information. Unlike the traditional generative models, the proposed Sub-GAN can control the diversity of the generated samples via the multiplicity of the learned subspaces. Moreover, the Sub-GAN follows an unsupervised fashion to explore not only the visual classes but the latent continuous attributes. We demonstrate that our model can discover meaningful visual attributes which is hard to be annotated via strong supervision, e.g., the writing style of digits, thus avoid the mode collapse problem. Extensive experimental results show the competitive performance of the proposed method for both generating diverse images with satisfied quality and discovering discriminative latent subspaces.

Local structure of target objects are essential for robust tracking. However, existing methods based on deep neural networks mostly describe the target appearance from the global view, leading to high sensitivity to non-rigid appearance change and partial occlusion. In this paper, we circumvent this issue by proposing a local structure learning method, which simultaneously considers the local patterns of the target and their structural relationships for more accurate target tracking. To this end, a local pattern detection module is designed to automatically identify discriminative regions of the target objects. The detection results are further refined by a message passing module, which enforces the structural context among local patterns to construct local structures. We show that the message passing module can be formulated as the inference process of a conditional random field (CRF) and implemented by differentiable operations, allowing the entire model to be trained in an end-to-end manner. By considering various combinations of the local structures, our tracker is able to form various types of structure patterns. Target tracking is finally achieved by a matching procedure of the structure patterns between target template and candidates. Extensive evaluations on three benchmark data sets demonstrate that the proposed tracking algorithm performs favorably against state-of-the-art methods while running at a highly efficient speed of 45 fps.

Many recent energy-based methods for optical flow estimation rely on a good initialization that is typically provided by some kind of feature matching. So far, however, these initial matching approaches are rather general: They do not incorporate any additional information that could help to improve the accuracy or the robustness of the estimation. In particular, they do not exploit potential cues on the camera poses and the thereby induced rigid motion of the scene. In the present paper, we tackle this problem. To this end, we propose a novel structure-from-motion-aware PatchMatch approach that, in contrast to existing matching techniques, combines two hierarchical feature matching methods: a recent two-frame PatchMatch approach for optical flow estimation (general motion) and a specifically tailored three-frame PatchMatch approach for rigid scene reconstruction (SfM). While the motion PatchMatch serves as baseline with good accuracy, the SfM counterpart takes over at occlusions and other regions with insufficient information. Experiments with our novel SfM-aware PatchMatch approach demonstrate its usefulness. They not only show excellent results for all major benchmarks (KITTI 2012/2015, MPI Sintel), but also improvements up to 50 percent compared to a PatchMatch approach without structure information.

Colorizing a given gray-level image is an important task in the media and advertising industry. Due to the ambiguity inherent to colorization (many shades are often plausible), recent approaches started to explicitly model diversity. However, one of the most obvious artifacts, structural inconsistency, is rarely considered by existing methods which predict chrominance independently for every pixel. To address this issue, we develop a conditional random field based variational auto-encoder formulation which is able to achieve diversity while taking into account structural consistency. Moreover, we introduce a controllability mecha- nism that can incorporate external constraints from diverse sources in- cluding a user interface. Compared to existing baselines, we demonstrate that our method obtains more diverse and globally consistent coloriza- tions on the LFW, LSUN-Church and ILSVRC-2015 datasets.

The Fast Style Transfer methods have been recently proposed to transfer a photograph to an artistic style in real-time. This task involves controlling the stroke size in the stylized results, which remains an open challenge. In this paper, we present a stroke controllable style transfer network that can achieve continuous and spatial stroke size control. By analyzing the factors that influence the stroke size, we propose to explicitly account for the receptive field and the style image scales. We propose a StrokePyramid module to endow the network with adaptive receptive fields, and two training strategies to achieve faster convergence and augment new stroke sizes upon a trained model respectively. By combining the proposed runtime control strategies, our network can achieve continuous changes in stroke sizes and produce distinct stroke sizes in different spatial regions within the same output image.

Question answering is an important task for autonomous agents and virtual assistants alike and was shown to support the disabled in efficiently navigating an overwhelming environment. Many existing methods focus on observation-based questions, ignoring our ability to seamlessly combine observed content with general knowledge. To understand interactions with a knowledge base, a dataset has been introduced recently and keyword matching techniques were shown to yield compelling results despite being vulnerable to misconceptions due to synonyms and homographs. To address this issue, we develop a learning-based approach which goes straight to the facts via a learned embedding space. We demonstrate state-of-the-art results on the challenging recently introduced fact-based visual question answering dataset, outperforming competing methods by more than 5%.

This paper presents StereoNet, the first end-to-end deep architecture for real-time stereo matching that runs at 60 fps on an NVidia Titan X, producing high-quality, edge-preserved, quantization-free depth maps. A key insight of this paper is that the network achieves a sub-pixel matching precision than is a magnitude higher than those of traditional stereo matching approaches. This allows us to achieve real-time performance by using a very low resolution cost volume that encodes all the information needed to achieve high depth precision. Spatial precision is achieved by employing a learned edge-aware upsampling function. Our model uses a Siamese network to extract features from the left and right image. A first estimate of the disparity is computed in a very low resolution cost volume, then hierarchically the model re-introduces high-frequency details through a learned upsampling function that uses compact pixel-to-pixel refinement networks. Leveraging color input as a guide, this function is capable of producing high-quality edge-aware output. We achieve compelling results on multiple benchmarks, showing how the proposed method offers extreme flexibility at an acceptable computational budget.

Stereo relative pose problem lies at the core of stereo visual odometry systems that are used in many applications. In this work we present two minimal solvers for stereo relative pose. We specifically con- sider the case when a minimal set consist of three point or line features and each of them has three known projections on two stereo cameras. We validate the importance of this formulation for practical purposes in our experiments with motion estimation. We then present a complete classi- fication of minimal cases with three point or line correspondences each having three projections, and present two new solvers that can handle all such cases. We demonstrate a considerable effect from the integration of the new solvers into a visual SLAM system.

We propose a stereo vision-based approach for tracking the camera ego-motion and 3D semantic objects in dynamic autonomous driving scenarios. Instead of directly regressing the 3D bounding box using end-to-end approaches, we propose to use the easy-to-labeled 2D detection and discrete viewpoint classification together with a light-weight semantic inference method to obtain rough 3D object measurements. Based on the object-aware-aided camera pose tracking which is robust in dynamic environments, in combination with our novel dynamic object bundle adjustment (BA) approach to fuse temporal sparse feature correspondences and the semantic 3D measurement model, we obtain 3D object pose, velocity and anchored dynamic point cloud estimation with instance accuracy and temporal consistency. The performance of our proposed method is demonstrated in diverse scenarios. Both the ego-motion estimation and object localization are compared with the state-of-of-the-art solutions.

This paper proposes an original problem of emph{stereo computation from a single (additive) mixture image}-- a challenging problem that had not been researched before. The goal is to separate (ie unmix) a single mixture image into two constitute image layers, such that the two layers form a left-right stereo image pair, from which a valid disparity map can be recovered. This is a severely illposed problem, from one input image one effectively aims to recover three (ie, left image, right image and a disparity map). In this work we give a novel deep-learning based solution, by jointly solving the two subtasks of image layer separation as well as stereo matching. Training our deep net is a simple task, as it does not need to have disparity maps. Extensive experiments demonstrate the efficacy of our method.