We investigate conditional adversarial networks as a general-purpose solution to image-to-image translation problems. These networks not only learn the mapping from input image to output image, but also learn a loss function to train this mapping. This makes it possible to apply the same generic approach to problems that traditionally would require very different loss formulations. We demonstrate that this approach is effective at synthesizing photos from label maps, reconstructing objects from edge maps, and colorizing images, among other tasks. Moreover, since the release of the pix2pix software associated with this paper, hundreds of twitter users have posted their own artistic experiments using our system. As a community, we no longer hand-engineer our mapping functions, and this work suggests we can achieve reasonable results without handengineering our loss functions either.

/pdf/image-to-image-translation-with-conditional-adversarial-5058t24et2.pdf

Image-to-Image Translation with Conditional Adversarial Networks

We investigate conditional adversarial networks as a general-purpose solution to image-to-image translation problems. These networks not only learn the mapping from input image to output image, but also learn a loss function to train this mapping. This makes it possible to apply the same generic approach to problems that traditionally would require very different loss formulations. We demonstrate that this approach is effective at synthesizing photos from label maps, reconstructing objects from edge maps, and colorizing images, among other tasks. Indeed, since the release of the pix2pix software associated with this paper, a large number of internet users (many of them artists) have posted their own experiments with our system, further demonstrating its wide applicability and ease of adoption without the need for parameter tweaking. As a community, we no longer hand-engineer our mapping functions, and this work suggests we can achieve reasonable results without hand-engineering our loss functions either.

3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representation automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet - a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.

/pdf/3d-shapenets-a-deep-representation-for-volumetric-shapes-1ii3phcu91.pdf

3D ShapeNets: A deep representation for volumetric shapes

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Automatic estimation of salient object regions across images, without any prior assumption or knowledge of the contents of the corresponding scenes, enhances many computer vision and computer graphics applications. We introduce a regional contrast based salient object detection algorithm, which simultaneously evaluates global contrast differences and spatial weighted coherence scores. The proposed algorithm is simple, efficient, naturally multi-scale, and produces full-resolution, high-quality saliency maps. These saliency maps are further used to initialize a novel iterative version of GrabCut, namely SaliencyCut, for high quality unsupervised salient object segmentation. We extensively evaluated our algorithm using traditional salient object detection datasets, as well as a more challenging Internet image dataset. Our experimental results demonstrate that our algorithm consistently outperforms 15 existing salient object detection and segmentation methods, yielding higher precision and better recall rates. We also show that our algorithm can be used to efficiently extract salient object masks from Internet images, enabling effective sketch-based image retrieval (SBIR) via simple shape comparisons. Despite such noisy internet images, where the saliency regions are ambiguous, our saliency guided image retrieval achieves a superior retrieval rate compared with state-of-the-art SBIR methods, and additionally provides important target object region information.

https://mmcheng.net/mftp/SalObj/PosterSaliency.pdf

Global contrast based salient region detection

General image completion and extrapolation methods often fail on portrait images where parts of the human body need to be recovered - a task that requires accurate human body structure and appearance synthesis. We present a two-stage deep learning framework for tacking this problem. In the first stage, given a portrait image with an incomplete human body, we extract a complete, coherent human body structure through a human parsing network, which focuses on structure recovery inside the unknown region with the help of pose estimation. In the second stage, we use an image completion network to fill the unknown region, guided by the structure map recovered in the first stage. For realistic synthesis the completion network is trained with both perceptual loss and conditional adversarial loss. We evaluate our method on public portrait image datasets, and show that it outperforms other state-of-art general image completion methods. Our method enables new portrait image editing applications such as occlusion removal and portrait extrapolation. We further show that the proposed general learning framework can be applied to other types of images, e.g. animal images.

Deep Portrait Image Completion and Extrapolation

We propose a localization refinement approach for candidate traffic signs. Previous traffic sign localization approaches, which place a bounding rectangle around the sign, do not always give a compact bounding box, making the subsequent classification task more difficult. We formulate localization as a segmentation problem, and incorporate prior knowledge concerning color and shape of traffic signs. To evaluate the effectiveness of our approach, we use it as an intermediate step between a standard traffic sign localizer and a classifier. Our experiments use the well-known German Traffic Sign Detection Benchmark (GTSDB) as well as our new Chinese Traffic Sign Detection Benchmark. This newly created benchmark is publicly available, 1 and goes beyond previous benchmark data sets: it has over 5000 high-resolution images containing more than 14 000 traffic signs taken in realistic driving conditions. Experimental results show that our localization approach significantly improves bounding boxes when compared with a standard localizer, thereby allowing a standard traffic sign classifier to generate more accurate classification results. 1 http://cg.cs.tsinghua.edu.cn/ctsdb/

/pdf/an-optimization-approach-for-localization-refinement-of-2fl7qy8nr7.pdf

An Optimization Approach for Localization Refinement of Candidate Traffic Signs

Compared to 2D textures, solid textures can represent not only the bounding surfaces, but also their interiors. Existing solid texture synthesis methods pay little attention to the generation of conforming textures that capture geometric structures or reflect the artists’ design intentions. In this paper, we propose a novel approach to synthesizing solid textures using 2D exemplars. The generated textures locally agree with a tensor field derived from user sketching curves. We use a deterministic approach and only a small portion of the voxels needs to be synthesized on demand. Correction is fundamental in deterministic texture synthesis. We propose a history windows representation, which is general enough to unifiedly represent various previous correction schemes, and a dual grid scheme based on it to significantly reduce the dependent voxels while still producing high quality results. Experiments demonstrate that our method produces significantly improved solid textures with a small amount of user interaction.

/pdf/sketch-guided-solid-texturing-4nvlna85as.pdf

Sketch guided solid texturing

Automultiscopic screens present different images depending on the viewing direction. This enables glasses-free 3D and provides motion parallax effect. However, due to the limited angular resolution of such displays, they suffer from hot-spotting, i. e., image quality is highly affected by the viewing position. In this paper, we analyze light fields produced by lenticular and parallax-barrier displays, and show that, unlike in real world, the light fields produced by such screens have a repetitive structure. This induces visual artifacts in the form of view discontinuities, depth reversals, and excessive disparities when viewing position is not optimal. Although the problem has been always considered as inherent to the technology, we demonstrate that light fields reproduced on automultiscopic displays have enough degrees of freedom to improve the visual quality. We propose a new technique that modifies light fields using global and local shears followed by stitching to improve their continuity when displayed on a screen. We show that this enhances visual quality significantly, which is demonstrated in a series of user experiments with an automultiscopic display as well as lenticular prints.

/pdf/improving-visual-quality-of-view-transitions-in-zaq8wr2k2j.pdf

Improving visual quality of view transitions in automultiscopic displays

A new type of bivariate generalized Ball basis function on a triangle is presented for free-form surface design. Some properties of the basis function are given, then degree elevation, recursive evaluation and some other properties of the generalized Ball surfaces are also derived. It is shown that the proposed recursive evaluation algorithm is more efficient than those of the old surfaces.

Shi-Min Hu

Papers

Deep Portrait Image Completion and Extrapolation

An Optimization Approach for Localization Refinement of Candidate Traffic Signs

Sketch guided solid texturing

Improving visual quality of view transitions in automultiscopic displays

A type of triangular ball surface and its properties