In this work, we connect two distinct concepts for unsupervised domain adaptation: feature distribution alignment between domains by utilizing the task-specific decision boundary [58] and the Wasserstein metric [73]. Our proposed sliced Wasserstein discrepancy (SWD) is designed to capture the natural notion of dissimilarity between the outputs of task-specific classifiers. It provides a geometrically meaningful guidance to detect target samples that are far from the support of the source and enables efficient distribution alignment in an end-to-end trainable fashion. In the experiments, we validate the effectiveness and genericness of our method on digit and sign recognition, image classification, semantic segmentation, and object detection.

Abstract. We present a tree-structured network architecture for largescale image classification. The trunk of the network contains convolutional layers optimized over all classes. At a given depth, the trunk splits into separate branches, each dedicated to discriminate a different subset of classes. Each branch acts as an expert classifying a set of categories that are difficult to tell apart, while the trunk provides common knowledge to all experts in the form of shared features. The training of our "network of experts" is completely end-to-end: the partition of categories into disjoint subsets is learned simultaneously with the parameters of the network trunk and the experts are trained jointly by minimizing a single learning objective over all classes. The proposed structure can be built from any existing convolutional neural network (CNN). We demonstrate its generality by adapting 4 popular CNNs for image categorization into the form of networks of experts. Our experiments on CIFAR100 and ImageNet show that in every case our method yields a substantial improvement in accuracy over the base CNN, and gives the best result achieved so far on CIFAR100. Finally, the improvement in accuracy comes at little additional cost: compared to the base network, the training time is only moderately increased and the number of parameters is comparable or in some cases even lower. Our code is available at:

Most nonrigid objects exhibit temporal regularities in their deformations. Recently it was proposed that these regularities can be parameterized by assuming that the nonrigid structure lies in a small dimensional trajectory space. In this paper, we propose a factorization approach for 3D reconstruction from multiple static cameras under the compact trajectory subspace representation. Proposed factorization is analogous to rank-3 factorization of rigid structure from motion problem, in transformed space. The benefit of our approach is that the 3D trajectory basis can be directly learned from the image observations. This also allows us to impute missing observations and denoise tracking errors without explicit estimation of the 3D structure. In contrast to standard triangulation based methods which require points to be visible in at least two cameras, our approach can reconstruct points, which remain occluded even in all the cameras for quite a long time. This makes our solution especially suitable for occlusion handling in motion capture systems. We demonstrate robustness of our method on challenging real and synthetic scenarios.

In this paper we propose a method for estimating depth from a single image using a coarse to fine approach. We argue that modeling the fine depth details is easier after a coarse depth map has been computed. We express a global (coarse) depth map of an image as a linear combination of a depth basis learned from training examples. The depth basis captures spatial and statistical regularities and reduces the problem of global depth estimation to the task of predicting the input-specific coefficients in the linear combination. This is formulated as a regression problem from a holistic representation of the image. Crucially, the depth basis and the regression function are coupled and jointly optimized by our learning scheme. We demonstrate that this results in a significant improvement in accuracy compared to direct regression of depth pixel values or approaches learning the depth basis disjointly from the regression function. The global depth estimate is then used as a guidance by a local refinement method that introduces depth details that were not captured at the global level. Experiments on the NYUv2 and KITTI datasets show that our method outperforms the existing state-of-the-art at a considerably lower computational cost for both training and testing.

No abstract