「CV」 姿态识别资源汇总
pose estimation
· action classification
又叫动作识别,给定一幅图像或一段视频,人体姿态识别就是去恢复其中人体关节点位置的过程;
当然,后来也发展出,根据已有的姿态序列,预测下一个姿态;
视频动作识别资源
1 综述
2 理论
- Unifying Few- and Zero-Shot Egocentric Action Recognition
CVPR 2020 2020-05-27 paper
开集动作识别——少样本;
3 识别
3.1 头部
- Improving Head Pose Estimation with a Combined Loss and Bounding Box Margin Adjustment
2019-05-14 paper
3.2 身体
- Being the center of attention: A Person-Context CNN framework for Personality Recognition
2010-10-15 paper
从姿态看社会属性;
3.3 3D
- Patch-based 3D Human Pose Refinement
CVPR 2019 (Augmented Human) 2019-05-20 paper
4 预测
- Trajectorylet-Net: a novel framework for pose prediction based on trajectorylet descriptors
2019-10-15 paper
Trajectorylet-Net;
附录
A 参考资料
- 程程. VALSE2017系列之三:人体姿态识别领域年度进展报告[EB/OL]. https://zhuanlan.zhihu.com/p/27293180. 2017-06-07/2019-05-14.
- lyl-Deepleraing. 深度学习的人体动作识别算法[EB/OL]. https://zhuanlan.zhihu.com/p/46599638. 2018-10-12/2019-05-14.
- 人体姿态识别研究综述
- 姿态识别综述
- 姿态估计与行为识别(行为检测、行为分类)的区别
Comments