CV Code | 本周新出计算机视觉开源代码汇总（含目标跟踪、语义分割、姿态跟踪、少样本学习等）

刚刚过去的一周出现了很多很实用、有意思、很神奇的CV代码。

比如大家期待的SiamRPN++算法，官方终于要开源了。

阿里MNN成为移动端网络部署的新选择。

同时CVPR 2019的论文也有几篇开源了，其中还有一篇做难民识别，原来计算机视觉也可用于直接解决社会问题。

本周曾经跟大家解读过 重磅！MobileNetV3 来了！，可惜谷歌并没有开源，但下面的另一篇轻量级网络设计的文章也值得大家参考哦。

说到轻量级，轻量级实时语义分割网络LEDNet，你一定不要错过！

来自英伟达的“仅需少量样本非监督的图像转换”的FUNIT算法，吸引了不少人的围观，这更加接近了算法应用了，也许会催生出一批好玩的应用。

一起来看看吧～

商汤将开源目标跟踪研究平台PySOT，含目前最强大的跟踪算法SiamRPN++

https://github.com/STVIR/pysot

阿里巴巴开源轻量级深度神经网络推理引擎MNN

https://github.com/alibaba/MNN

CVPR 2019

深度学习用于线段检测

PPGNet: Learning Point-Pair Graph for Line Segment Detection

Ziheng Zhang, Zhengxin Li, Ning Bi, Jia Zheng, Jinlei Wang, Kun Huang, Weixin Luo, Yanyu Xu, Shenghua Gao

https://arxiv.org/abs/1905.03415v1

https://github.com/svip-lab/PPGNet

（还未放出源码）

非监督学习用于内容感知的图像重定向

（图像重定向是指不引入不可接受的畸变情况下改变图像大小和长宽比，比如用于适应不同分辨率屏幕）

Cycle-IR: Deep Cyclic Image Retargeting

Weimin Tan, Bo Yan, Chumin Lin, Xuejing Niu

https://arxiv.org/abs/1905.03556v1

https://github.com/mintanwei/Cycle-IR

行人重识别

Frustratingly Easy Person Re-Identification: Generalizing Person Re-ID in Practice

Jieru Jia, Qiuqi Ruan, Timothy M. Hospedales

https://arxiv.org/abs/1905.03422v1

（将开源，还未放出地址）

轻量级网络设计，重新思考逆残差结构

Seesaw-Net: Convolution Neural Network With Uneven Group Convolution

Jintao Zhang

https://arxiv.org/abs/1905.03672v1

（将开源，还未放出地址）

CVPR 2019

用于语音驱动的3D人脸动画的合成数据集与模型

Capture, Learning, and Synthesis of 3D Speaking Styles

Daniel Cudeiro, Timo Bolkart, Cassidy Laidlaw, Anurag Ranjan, Michael J. Black

https://arxiv.org/abs/1905.03079v1

http://voca.is.tue.mpg.de/

端到端框架解析，精度大幅提高

End-to-End Wireframe Parsing

Yichao Zhou, Haozhi Qi, Yi Ma

https://arxiv.org/abs/1905.03246v1

https://github.com/zhou13/lcnn

姿态跟踪框架

LightTrack: A Generic Framework for Online Top-Down Human Pose Tracking

Guanghan Ning, Heng Huang

https://arxiv.org/abs/1905.02822v1

https://github.com/Guanghan/lighttrack

3D多目标跟踪，特征关联网络

IEEE Intelligent Vehicles Symposium (IV 19)

FANTrack: 3D Multi-Object Tracking with Feature Association Network

Erkan Baser, Venkateshwaran Balasubramanian, Prarthana Bhattacharyya, Krzysztof Czarnecki

https://arxiv.org/abs/1905.02843v1

https://git.uwaterloo.ca/wise-lab/fantrack

集成使用不同激活函数训练的神经网络，取得更高的精度

Ensemble of Convolutional Neural Networks Trained with Different Activation Functions

Gianluca Maguolo, Loris Nanni, Stefano Ghidoni

https://arxiv.org/abs/1905.02473v1

https://github.com/LorisNanni

增强可变形卷积神经网络用于视频修复

CVPR 2019 Workshop

EDVR: Video Restoration with Enhanced Deformable Convolutional Networks

Xintao Wang, Kelvin C.K. Chan, Ke Yu, Chao Dong, Chen Change Loy

https://arxiv.org/abs/1905.02716v1

https://github.com/xinntao/EDVR

复杂室内场景的逆渲染

Inverse Rendering for Complex Indoor Scenes: Shape, Spatially-Varying Lighting and SVBRDF from a Single Image

Zhengqin Li, Mohammad Shafiei, Ravi Ramamoorthi, Kalyan Sunkavalli, Manmohan Chandraker

https://arxiv.org/abs/1905.02722v1

（将开源，还未放出地址）

大规模动态环境额鲁棒密度匹配

ICRA 2018

Robust Dense Mapping for Large-Scale Dynamic Environments

Ioan Andrei Bârsan, Peidong Liu, Marc Pollefeys, Andreas Geiger

https://arxiv.org/abs/1905.02781v1

http://andreibarsan.github.io/dynslam

无监督多视图立体视觉，使用鲁棒光照一致性方法

Learning Unsupervised Multi-View Stereopsis via Robust Photometric Consistency

Tejas Khot, Shubham Agrawal, Shubham Tulsiani, Christoph Mertz, Simon Lucey, Martial Hebert

https://arxiv.org/abs/1905.02706v1

https://tejaskhot.github.io/unsup_mvs/

ICIP 2019

轻量级实时语义分割网络

LEDNet: A Lightweight Encoder-Decoder Network for Real-Time Semantic Segmentation

Yu Wang, Quan Zhou, Jia Liu, Jian Xiong, Guangwei Gao, Xiaofu Wu, Longin Jan Latecki

https://arxiv.org/abs/1905.02423v1

https://github.com/xiaoyufenfei/LEDNet

单幅图像3D手部重建

Single Image 3D Hand Reconstruction with Mesh Convolutions

Dominik Kulon, Haoyang Wang, Riza Alp Güler, Michael Bronstein, Stefanos Zafeiriou

https://arxiv.org/abs/1905.01326v1

（将开源，还未放出地址）

使用贝叶斯优化学习最优的数据增广策略，针对图像分类任务

Learning Optimal Data Augmentation Policies via Bayesian Optimization for Image Classification Tasks

Chunxu Zhang, Jiaxu Cui, Bo Yang

https://arxiv.org/abs/1905.02610v1

https://github.com/zhangxiaozao/BO-Aug

仅需少量样本非监督的图像转换

Few-Shot Unsupervised Image-to-Image Translation

Ming-Yu Liu, Xun Huang, Arun Mallya, Tero Karras, Timo Aila, Jaakko Lehtinen, Jan Kautz

https://arxiv.org/abs/1905.01723v1

https://nvlabs.github.io/FUNIT

CVPR 2019 Workshop on Computer Vision for Global Challenges (CV4GC)

难民识别

DisplaceNet: Recognising Displaced People from Images by Exploiting Dominance Level

Grigorios Kalliatakis, Shoaib Ehsan, Maria Fasli, Klaus McDonald-Maier

https://arxiv.org/abs/1905.02025v1

https://github.com/GKalliatakis/DisplaceNet

CVPR 2019 Oral

少样本学习，使用图神经网络去噪自动编码机生成分类权重

Generating Classification Weights with GNN Denoising Autoencoders for Few-Shot Learning

Spyros Gidaris, Nikos Komodakis

https://arxiv.org/abs/1905.01102v1

https://github.com/gidariss/wDAE_GNN_FewShot