Attention-A-Lightweight-2D-Hand-Pose-Estimation-Approach-Pytorch

TensorFlow 版本：https://github.com/hanchenchen/Attention-A-Lightweight-2D-Hand-Pose-Estimation-Approach-master/tree/dev

主要任务

在手部关键点检测任务中，使用论文 Attention! A Lightweight 2D Hand Pose Estimation Approach 中提出的Attention Augmented Inverted Bottleneck Block等结构。

测评环境

• Ubuntu 16.04.6 LTS
• Python 3.9.1
• Cuda 10.1

Update

1. corrected the number of parameters
2. Download the pre-processed dataset according to: https://github.com/HowieMa/NSRMhand.

   mv path/to/CMUhand.tar ./
   tar -xvf CMUhand.tar
   

Reference

[1] Santavas N, Kansizoglou I, Bampis L, et al. Attention! a lightweight 2d hand pose estimation approach[J]. IEEE Sensors Journal, 2020. [[code]][https://github.com/nsantavas/Attention-A-Lightweight-2D-Hand-Pose-Estimation-Approach]

[2] Chen Y, Ma H, Kong D, et al. Nonparametric structure regularization machine for 2D hand pose estimation[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2020: 381-390. [code][https://github.com/HowieMa/NSRMhand] (This project is manually forked from this project. )

[3] Wei S E, Ramakrishna V, Kanade T, et al. Convolutional pose machines[C]//Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. 2016: 4724-4732.

[4] Simon T, Joo H, Matthews I, et al. Hand keypoint detection in single images using multiview bootstrapping[C]//Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. 2017: 1145-1153. [Panoptic][http://domedb.perception.cs.cmu.edu/handdb.html]

[5] Zimmermann C, Ceylan D, Yang J, et al. Freihand: A dataset for markerless capture of hand pose and shape from single rgb images[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2019: 813-822. [FreiHAND][https://lmb.informatik.uni-freiburg.de/projects/freihand/]

[6] Zhang J, Jiao J, Chen M, et al. 3d hand pose tracking and estimation using stereo matching[J]. arXiv preprint arXiv:1610.07214, 2016. [SHP]

[7] Shivakumar S H, Oberweger M, Rad M, et al. HO-3D: A Multi-User, Multi-Object Dataset for Joint 3D Hand-Object Pose Estimation[J]. arXiv. org e-Print archive, 2019. [HO3D_v2]