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

Point clouds provide a flexible geometric representation suitable for countless applications in computer graphics; they also comprise the raw output of most 3D data acquisition devices. While hand-designed features on point clouds have long been proposed in graphics and vision, however, the recent overwhelming success of convolutional neural networks (CNNs) for image analysis suggests the value of adapting insight from CNN to the point cloud world. Point clouds inherently lack topological information, so designing a model to recover topology can enrich the representation power of point clouds. To this end, we propose a new neural network module dubbed EdgeConv suitable for CNN-based high-level tasks on point clouds, including classification and segmentation. EdgeConv acts on graphs dynamically computed in each layer of the network. It is differentiable and can be plugged into existing architectures. Compared to existing modules operating in extrinsic space or treating each point independently, EdgeConv has several appealing properties: It incorporates local neighborhood information; it can be stacked applied to learn global shape properties; and in multi-layer systems affinity in feature space captures semantic characteristics over potentially long distances in the original embedding. We show the performance of our model on standard benchmarks, including ModelNet40, ShapeNetPart, and S3DIS.

/pdf/dynamic-graph-cnn-for-learning-on-point-clouds-20wlcru9tg.pdf

Dynamic Graph CNN for Learning on Point Clouds

On robust estimation of the location parameter

The PASCAL Visual Object Classes Challenge

One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs (SN-GANs) is capable of generating images of better or equal quality relative to the previous training stabilization techniques.

/pdf/spectral-normalization-for-generative-adversarial-networks-3dxutlj1bi.pdf

Spectral Normalization for Generative Adversarial Networks

Existing volumetric capture systems require many cameras and lengthy post processing. We introduce the first system that can capture a completely clothed human body (including the back) using a single RGB webcam and in real time. Our deep-learning-based approach enables new possibilities for low-cost and consumeraccessible immersive teleportation.

Volumetric human teleportation

Many mission-critical systems are based on GPU for inference It requires not only high recognition accuracy but also low latency in responding time Although many studies are devoted to optimizing the structure of deep models for efficient inference, most of them do not leverage the architecture of \textbf{modern GPU} for fast inference, leading to suboptimal performance To address this issue, we propose a general principle for designing GPU-efficient networks based on extensive empirical studies This design principle enables us to search for GPU-efficient network structures effectively by a simple and lightweight method as opposed to most Neural Architecture Search (NAS) methods that are complicated and computationally expensive Based on the proposed framework, we design a family of GPU-Efficient Networks, or GENets in short We did extensive evaluations on multiple GPU platforms and inference engines While achieving $\geq 813\%$ top-1 accuracy on ImageNet, GENet is up to $64$ times faster than EfficienNet on GPU It also outperforms most state-of-the-art models that are more efficient than EfficientNet in high precision regimes Our source code and pre-trained models are available from \url{this https URL}

Neural Architecture Design for GPU-Efficient Networks

Recently, GAN based method has demonstrated strong effectiveness in generating augmentation data for person re-identification (ReID), on account of its ability to bridge the gap between domains and enrich the data variety in feature space. However, most of the ReID works pick all the GAN generated data as additional training samples or evaluate the quality of GAN generation at the entire data set level, ignoring the image-level essential feature of data in ReID task. In this paper, we analyze the in-depth characteristics of ReID sample and solve the problem of "What makes a GAN-generated image good for ReID''. Specifically, we propose to examine each data sample with id-consistency and diversity constraints by mapping image onto different spaces. With a metric-based sampling method, we demonstrate that not every GAN-generated data is beneficial for augmentation. Models trained with data filtered by our quality evaluation outperform those trained with the full augmentation set by a large margin. Extensive experiments show the effectiveness of our method on both supervised ReID task and unsupervised domain adaptation ReID task.

https://dl.acm.org/doi/pdf/10.1145/3474085.3475547

Exploring the Quality of GAN Generated Images for Person Re-Identification

Background

Objective quantification of anatomical variations about the femur head–neck junction in pre-operative planning for surgical intervention in femoro-acetabular impingement is problematic, as no clear definition of average normal anatomy for a specific subject exists.



Methods

We have defined the normal-equivalent of a subject's anatomy by using a statistical shape model and geometric shape optimization for finding correspondences, while excluding the femoral head–neck junction during the fitting procedure. The presented technique was evaluated on a cohort of 20 patients.



Results

Difference in α-angle measurement between the actual morphology and the predicted normal-equivalent, averaged 1.3° (SD 1.7°) in the control group versus 8° (SD 7.3°) in the patient group (p < 0.05).



Conclusions

Defining normal equivalent anatomy is effective in quantifying anatomical dysmorphism of the femoral head–neck junction and as such can improve presurgical analysis of patients diagnosed with femoro-acetabular impingement. Copyright © 2016 John Wiley & Sons, Ltd.

Patient-specific assessment of dysmorphism of the femoral head–neck junction: a statistical shape model approach

Throughout the last few years, acquisition and processing of dynamic geometry has already received quite an amount of attention in the computer vision and graphics research community. Recently, the topic has gained a significant boost due to the availability of commodity devices for dynamic geometry acquisition: The introduction of the Microsoft ŞKinectŤ device made this kind of technology broadly available, being very well received by both researchersand end-users, and even more development in this direction can probably be expected for the near future. The tutorial on ŞDynamic Geometry ProcessingŤ considers the problem of processing such dynamic range data effectively and efficiently. The tutorial introduces basic processing techniques for analyzing and matching range data. It introduces models for correspondence estimation and presents the according basic algorithmic building blocks. Furthermore, it discusses the current state-of-the-art by looking at example approaches for processing and real-time tracking of dynamic data. In addition, the tutorial will also identify and discuss future challenges in the field, aiming at inspiring future work in this exciting area of research. c © The Eurographics Association 2012.

Hao Li

Papers

Volumetric human teleportation

Neural Architecture Design for GPU-Efficient Networks

Exploring the Quality of GAN Generated Images for Person Re-Identification

Patient-specific assessment of dysmorphism of the femoral head–neck junction: a statistical shape model approach

Dynamic Geometry Processing