目前我的准确率和速度还在一直提高,目前内部最新版本达到72mAP(调试中)。最新结果,我们会及时在Github和arvix上及时更新。
关于MVIG
我们研究组刚刚(2016年九月)成立,由上海交大卢策吾教授带领,主要研究计算机视觉,深度增强学习,深度学习
Homepage:https://cvsjtu.wordpress.com/
Group Github: https://github.com/MVIG-SJTU/
Prof. Cewu Lu is leading Machine Vision and Intelligent group at Shanghai Jiaotong University. He was a research fellow at Stanford University AI lab (under Fei-Fei Li and Leonidas Guibas) . He was a Research Assistant Professor at Hong Kong University Science and Technology(with CK Tang). He got Ph.D degree from The Chinese University of Hong Kong (Supervisor Jiaya Jia). His research interests include computer vision, deep reinforcement learning and deep learning.
[1] http://human-pose.mpi-inf.mpg.de/
[2] Eldar Insafutdinov, Leonid Pishchulin and Bjoern Andres, Mykhaylo Andriluka and Bernt Schiele DeeperCut: A Deeper, Stronger, and Faster Multi-Person Pose Estimation Model, European Conference on Computer Vision (ECCV),2016
项目介绍主页请点击“阅读原文”