ECCV 2018 论文列表

We present an approach to robust estimation of fundamental matrices from noisy data contaminated by outliers. The problem is cast as a series of weighted homogeneous least-squares problems, where robust weights are estimated using deep networks. The presented formulation acts directly on putative correspondences and thus fits into standard 3D vision pipelines that perform feature extraction, matching, and model fitting. The approach can be trained end-to-end and yields computationally efficient robust estimators. Our experiments indicate that the presented approach is able to train robust estimators that outperform classic approaches on real data by a significant margin.

State-of-the-art object detectors usually learn multi-scale representations to get better results by employing feature pyramids. However, the current designs for feature pyramids are still inefficient to integrate the semantic information over different scales. In this paper, we begin by investigating current feature pyramids solutions, and then reformulate the feature pyramid construction as the feature reconfiguration process. Finally, we propose a novel reconfiguration architecture to combine low-level representations with high-level semantic features in a highly-nonlinear yet efficient way. In particular, our architecture which consists of global attention and local reconfigurations, is able to gather task-oriented features across different spatial locations and scales, globally and locally. Both the global attention and local reconfiguration are lightweight, in-place, and end-to-end trainable. Using this method in the basic SSD system, our models achieve consistent and significant boosts compared with the original model and its other variations, without losing real-time processing speed.

We propose Deep Feature Factorization (DFF), a method capable of localizing similar semantic concepts within an image or a set of images. We use DFF to gain insight into a deep convolutional neural network's learned features, where we detect hierarchical cluster structures in feature space. This is visualized as heat maps, which highlight semantically matching regions across a set of images, revealing what the network `perceives' as similar. DFF can also be used to perform co-segmentation and co-localization, and we report state-of-the-art results on these tasks.

Modelling human free-hand sketches has become topical recently, driven by practical applications such as fine-grained sketch based image retrieval (FG-SBIR). Sketches are clearly related to photo edge-maps, but a human free-hand sketch of a photo is not simply a clean rendering of that photo's edge map. Instead there is a fundamental process of abstraction and iconic rendering, where overall geometry is warped and salient details are selectively included. In this paper we study this sketching process and attempt to invert it. We model this inversion by translating iconic free-hand sketches to contours that resemble more geometrically realistic projections of object boundaries, and separately factorise out the salient added details. This factorised re-representation makes it easier to match a free-hand sketch to a photo instance of an object. Specifically, we propose a novel unsupervised image style transfer model based on enforcing a cyclic embedding consistency constraint. A deep FG-SBIR model is then formulated to accommodate complementary discriminative detail from each factorised sketch for better matching with the corresponding photo. Our method is evaluated both qualitatively and quantitatively to demonstrate its superiority over a number of state-of-the-art alternatives for style transfer and FG-SBIR.

Efficient CNN designs like ResNets and DenseNet were proposed to improve accuracy vs efficiency trade-offs. They essentially increased the connectivity, allowing efficient information flow across layers. Inspired by these techniques, we propose to model connections between filters of a CNN using graphs which are simultaneously sparse and well connected. Sparsity results in efficiency while well connectedness can preserve the expressive power of the CNNs. We use a well-studied class of graphs from theoretical computer science that satisfies these properties known as Expander graphs. Expander graphs are used to model connections between filters in CNNs to design networks called X-Nets. We present two guarantees on the connectivity of X-Nets: Each node influences every node in a layer in logarithmic steps, and the number of paths between two sets of nodes is proportional to the product of their sizes. We also propose efficient training and inference algorithms, making it possible to train deeper and wider X-Nets effectively. Expander based models give a 4% improvement in accuracy on MobileNet over grouped convolutions, a popular technique, which has the same sparsity but worse connectivity. X-Nets give better performance trade-offs than the original ResNet and DenseNet-BC architectures. We achieve model sizes comparable to state-of-the-art pruning techniques using our simple architecture design, without any pruning. We hope that this work motivates other approaches to utilize results from graph theory to develop efficient network architectures.

Domain generalization aims to learn a classification model from multiple source domains and generalize it to unseen target domains. A critical problem in domain generalization involves learning domain-invariant representations. Let $X$ and $Y$ denote the features and the labels, respectively. Under the assumption that the conditional distribution $P(Y|X)$ remains unchanged across domains, earlier approaches to domain generalization learned the invariant representation $T(X)$ by minimizing the discrepancy of the marginal distribution $P(T(X))$. However, such an assumption of stable $P(Y|X)$ does not necessarily hold in practice. In addition, the representation learning function $T(X)$ is usually constrained to a simple linear transformation or shallow networks. To address the above two drawbacks, we propose an end-to-end conditional invariant deep domain generalization approach by leveraging deep neural networks for domain-invariant representation learning. The domain-invariance property is guaranteed through a conditional invariant adversarial network that can learn domain-invariant representations w.r.t. the joint distribution $P(T(X),Y)$ if the target domain data are not severely class unbalanced. We perform various experiments to demonstrate the effectiveness of the proposed method.

This paper presents a new deep learning approach for video-based scene classification. We design a Heterogeneous Deep Discriminative Model (HDDM) whose parameters are initialized by performing an unsupervised pre-training in a layer-wise fashion using Gaussian Restricted Boltzmann Machines (GRBM). In order to avoid the redundancy of adjacent frames, we extract spatiotemporal variation patterns within frames and represent them sparsely using Sparse Cubic Symmetrical Pattern (SCSP). Then, a pre-initialized HDDM is separately trained using the videos of each class to learn class-specific models. According to the minimum reconstruction error from the learnt class-specific models, a weighted voting strategy is employed for the classification. The performance of the proposed method is extensively evaluated on two action recognition datasets; UCF101 and Hollywood II, and three dynamic texture and dynamic scene datasets; DynTex, YUPENN, and Maryland. The experimental results and comparisons against state-of-the-art methods demonstrate that the proposed method consistently achieves superior performance on all datasets.

Modern deep learning systems successfully solve many perception tasks such as object pose estimation when the input image is of high quality. However, in challenging imaging conditions such as on low resolution images or when the image is corrupted by imaging artifacts, current systems degrade considerably in accuracy. While a loss in performance is unavoidable we would like our models to quantify their uncertainty in order to achieve robustness against images of varying quality. Probabilistic deep learning models combine the expressive power of deep learning with uncertainty quantification. In this paper, we propose a novel probabilistic deep learning model for the task of angular regression. Our model uses von Mises distributions to predict a distribution over object pose angle. Whereas a single von Mises distribution is making strong assumptions about the shape of the distribution, we extend the basic model to predict a mixture of von Mises distributions. We show how to learn a mixture model using a finite and infinite number of mixture components. Our model allow for likelihood-based training and efficient inference at test time. We demonstrate on a number of challenging pose estimation datasets that our model produces calibrated probability predictions and competitive or superior point estimates compared to the current state-of-the-art.

Due to the large cross-modality discrepancy between 2D sketches and 3D shapes, retrieving 3D shapes by sketches is a significantly challenging task. To address this problem, we propose a novel framework to learn a discriminative deep cross-modality adaptation model in this paper. Specifically, we first separately adopt two metric networks, following two deep convolutional neural networks (CNNs), to learn modality-specific discriminative features based on an importance-aware metric learning method. Subsequently, we explicitly introduce a cross-modality transformation network to compensate for the divergence between two modalities, which can transfer features of 2D sketches to the feature space of 3D shapes. We develop an adversarial learning based method to train the transformation model, by simultaneously enhancing the holistic correlations between data distributions of two modalities, and mitigating the local semantic divergences through minimizing a cross-modality mean discrepancy term. Experimental results on the SHREC 2013 and SHREC 2014 datasets clearly show the superior retrieval performance of our proposed model, compared to the state-of-the-art approaches.

The key point of image-text matching is how to accurately measure the similarity between visual and textual inputs. Despite the great progress of associating the deep cross-modal embeddings with the bi-directional ranking loss, developing the strategies for mining useful triplets and selecting appropriate margins remains a challenge in real applications. In this paper, we propose a cross-modal projection matching (CMPM) loss and a cross-modal projection classification (CMPC) loss for learning discriminative image-text embeddings. The CMPM loss minimizes the KL divergence between the projection compatibility distributions and the normalized matching distributions defined with all the positive and negative samples in a mini-batch. The CMPC loss attempts to categorize the vector projection of representations from one modality onto another with the improved norm-softmax loss, for further enhancing the feature compactness of each class. Extensive analysis and experiments on multiple datasets demonstrate the superiority of the proposed approach.

In this paper, we propose a novel 3D object detector that can exploit both LIDAR as well as cameras to perform very accurate localization. Towards this goal, we design an end-to-end learnable architecture that exploits continuous convolutions to fuse image and LIDAR feature maps at different levels of resolution. Our proposed continuous fusion layer encode both discrete-state image features as well as continuous geometric information. This enables us to design a novel, reliable and efficient end-to-end learnable 3D object detector based on multiple sensors. Our experimental evaluation on both KITTI as well as a large scale 3D object detection benchmark shows significant improvements over the state of the art.

Despite a lack of theoretical understanding, deep neural networks have achieved unparalleled performance in a wide range of applications. On the other hand, shallow representation learning with component analysis is associated with rich intuition and theory, but smaller capacity often limits its usefulness. To bridge this gap, we introduce Deep Component Analysis (DeepCA), an expressive multilayer model formulation that enforces hierarchical structure through constraints on latent variables in each layer. For inference, we propose a differentiable optimization algorithm implemented using recurrent Alternating Direction Neural Networks (ADNNs) that enable parameter learning using standard backpropagation. By interpreting feed-forward networks as single-iteration approximations of inference in our model, we provide both a novel perspective for understanding them and a practical technique for constraining predictions with prior knowledge. Experimentally, we demonstrate performance improvements on a variety of tasks, including single-image depth prediction with sparse output constraints.

In this paper, we study the problem of semi-supervised image recognition, which is to learn classifiers using both labeled and unlabeled images. We present Deep Co-Training, a deep learning based method inspired by the Co-Training framework. The original Co-Training learns two classifiers on two views which are data from different sources that describe the same instances. To extend this concept to deep learning, Deep Co-Training trains multiple deep neural networks to be the different views and exploits adversarial examples to encourage view difference, in order to prevent the networks from collapsing into each other. As a result, the co-trained networks provide different and complementary information about the data, which is necessary for the Co-Training framework to achieve good results. We test our method on SVHN, CIFAR-10/100 and ImageNet datasets, and our method outperforms the previous state-of-the-art methods by a large margin.

Clustering is a class of unsupervised learning methods that has been extensively applied and studied in computer vision. Little work has been done to adapt it to the end-to-end training of visual features on large-scale datasets. In this work, we present DeepCluster, a clustering method that jointly learns the parameters of a neural network and the cluster assignments of the resulting features. DeepCluster iteratively groups the features with a standard clustering algorithm, k-means, and uses the subsequent assignments as supervision to update the weights of the network. We apply DeepCluster to the unsupervised training of convolutional neural networks on large datasets like ImageNet and YFCC100M. The resulting model outperforms the current state of the art by a significant margin on all the standard benchmarks.

Noise is an inherent issue of low-light image capture, which is worsened on mobile devices due to their narrow apertures and small sensors. One strategy for mitigating noise in low-light situations is to increase the shutter time, allowing each photosite to integrate more light and decrease noise variance. However, there are two downsides of long exposures: (a) bright regions can exceed the sensor range, and (b) camera and scene motion will cause blur. Another way of gathering more light is to capture multiple short (thus noisy) frames in a “burst” and intelligently integrate the content, thus avoiding the above downsides. In this paper, we use the burst-capture strategy and implement the intelligent integration via a recurrent fully convolutional deep neural net (CNN). We build our novel, multi-frame architecture to be a simple addition to any single frame denoising model. The resulting architecture denoises all frames in a sequence of arbitrary length. We show that it achieves state of the art denoising results on our burst dataset, improving on the best published multi-frame techniques, such as VBM4D and FlexISP. Finally, we explore other applications of multi-frame image enhancement and show that our CNN architecture generalizes well to image super-resolution.

Boosting is a classic algorithm which has been successfully applied to diverse computer vision tasks. In the scenario of image denoising, however, the existing boosting algorithms are surpassed by the emerging learning-based models. In this paper, we propose a novel deep boosting framework (DBF) for denoising, which integrates several convolutional networks in a feed-forward fashion. Along with the integrated networks, however, the depth of the boosting framework is substantially increased, which brings difficulty to training. To solve this problem, we introduce the concept of dense connection that overcomes the vanishing of gradients during training. Furthermore, we propose a path-widening fusion scheme cooperated with the dilated convolution to derive a lightweight yet efficient convolutional network as the boosting unit, named Dilated Dense Fusion Network (DDFN). Comprehensive experiments demonstrate that our DBF outperforms existing methods on widely used benchmarks, in terms of different denoising tasks.

In this paper, we focus on exploring modality-temporal mutual information for RGB-D action recognition. In order to learn time-varying information and multi-modal features jointly, we propose a novel deep bilinear learning framework. In the framework, we propose bilinear blocks that consist of two linear pooling layers for pooling the input cube features from both modality and temporal directions, separately. To capture rich modality-temporal information and facilitate our deep bilinear learning, a new action feature called modality-temporal cube is presented in a tensor structure for characterizing RGB-D actions from a comprehensive perspective. Our method is extensively tested on two public datasets with four different evaluation settings, and the results show that the proposed method outperforms the state-of-the-art approaches.

We present a novel regularization approach to train neural networks that enjoys better generalization and test error than standard stochastic gradient descent. Our approach is based on the principles of cross-validation, where a validation set is used to limit the model overfitting. We formulate such principles as a bilevel optimization problem. This formulation allows us to define the optimization of a cost on the validation set subject to another optimization on the training set. The overfitting is controlled by introducing weights on each mini-batch in the training set and by choosing their values so that they minimize the error on the validation set. In practice, these weights define mini-batch learning rates in a gradient descent update equation that favor gradients with better generalization capabilities. Because of its simplicity, this approach can be integrated with other regularization methods and training schemes. We evaluate extensively our proposed algorithm on several neural network architectures and datasets, and find that it consistently improves the generalization of the model, especially when labels are noisy.

We present a method for simultaneously estimating 3D human pose and body shape from a sparse set of wide-baseline camera views. We train a symmetric convolutional autoencoder with a dual loss that enforces learning of a latent representation that encodes skeletal joint positions, and at the same time learns a deep representation for volumetric body shape. We harness the latter to up-scale input volumetric data by a factor of 4x, whilst recovering a 3D estimate of joint positions with equal or greater accuracy than the state of the art. Inference runs in real-time (25 fps) and has potential for passive human behavior monitoring where there is a requirement for high fidelity estimation of human body shape and pose.