Skip to content

Latest commit

 

History

History

configs

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
BRL
BRL
 
 
aagcn
aagcn
 
 
ctrgcn
ctrgcn
 
 
msg3d
msg3d
 
 
stgcn
stgcn
 
 
README.md
README.md
 
 

README.md

Configs

Introduction

This directory includes configs for different recognition algorithms. We provide the results trained on NTU RGB+D 60, NTU RGB+D 120 and Kinetics-400 in the long-tailed training setting. We also provide checkpoints for six modalities: Joint, Bone, Skip, Joint Motion, Bone Motion, and Skip Motion.

BRL Model Zoo

We release numerous checkpoints for BRL trained with various modalities, annotations on NTU RGB+D 60, NTU RGB+D 120 and Kinetics-400. The accuracy of each modality links to the weight file.

Dataset Annotation Joint Top1 Bone Top1 Skip Top1 Joint Motion Top1 Bone Motion Top1 Skip Motion Top1 Two Stream Top1 Four Stream Top1 Six Stream Top1
NTURGB+D XSub Official 3D Skeleton joint_config: 76.7 bone_config: 76.1 skip_config: 77.7 joint_motion_config: 75.0 bone_motion_config: 72.8 skip_motion_config: 73.4 79.6 81.0 81.8
NTURGB+D XView Official 3D Skeleton joint_config: 81.4 bone_config: 80.3 skip_config: 81.1 joint_motion_config: 78.5 bone_motion_config: 76.2 skip_motion_config: 77.2 84.0 84.9 85.4
NTURGB+D 120 XSub Official 3D Skeleton joint_config: 65.3 bone_config: 65.3 skip_config: 64.2 joint_motion_config: 59.7 bone_motion_config: 59.8 skip_motion_config: 59.6 68.7 69.4 69.7
NTURGB+D 120 XSet Official 3D Skeleton joint_config: 66.8 bone_config: 66.6 skip_config: 65.9 joint_motion_config: 63.5 bone_motion_config: 62.2 skip_motion_config: 61.6 69.7 71.0 71.3
Kinetics-400 HRNet 2D Pose joint_config: 45.6 bone_config: 45.2 joint_motion_config: 42.0 bone_motion_config: 41.2 48.1 (1:1) 48.6 (3:3:1:1)

Supported Algorithms

Other Algorithms

We test the performance of Shift-GCN, DC-GCN+ADG, MST-GCN, and InfoGCN on the NTURGB+D 60-LT and NTURGB+D 120-LT datasets.

Shift-GCN

Dataset Annotation Joint Top1 Bone Top1 Joint Motion Top1 Bone Motion Top1 Four Stream Top1
NTURGB+D XSub Skeleton 63.7 61.4 64.3 62.5 73.6
NTURGB+D XView Skeleton 71.3 69.7 68.2 68.8 79.3
NTURGB+D 120 XSub Skeleton 54.7 54.2 51.2 53.2 62.3
NTURGB+D 120 XSet Skeleton 55.4 54.4 54.6 55.1 64.5

DC-GCN+ADG

Dataset Annotation Joint Top1 Bone Top1 Joint Motion Top1 Bone Motion Top1 Four Stream Top1
NTURGB+D XSub Skeleton 68.9 65.1 61.1 60.5 75.0
NTURGB+D XView Skeleton 74.9 69.9 68.1 67.8 79.7
NTURGB+D 120 XSub Skeleton 56.1 55.5 50.1 51.2 63.4
NTURGB+D 120 XSet Skeleton 59.6 56.7 55.6 53.9 66.2

MST-GCN

Dataset Annotation Joint Top1 Bone Top1 Joint Motion Top1 Bone Motion Top1 Four Stream Top1
NTURGB+D XSub Skeleton 70.1 70.2 67.2 68.0 75.9
NTURGB+D XView Skeleton 76.7 75.6 73.2 72.9 80.3
NTURGB+D 120 XSub Skeleton 57.5 59.9 54.6 56.2 63.8
NTURGB+D 120 XSet Skeleton 61.6 60.6 58.2 59.0 65.9

InfoGCN

Dataset Annotation Joint Top1 Bone Top1 Joint Motion Top1 Bone Motion Top1 Four Stream Top1
NTURGB+D XSub Skeleton 69.7 66.6 68.5 67.5 76.8
NTURGB+D XView Skeleton 73.5 68.5 72.2 71.5 79.2
NTURGB+D 120 XSub Skeleton 55.6 60.4 52.3 52.1 64.2
NTURGB+D 120 XSet Skeleton 58.5 60.8 57.8 55.7 67.1

Citation

% ST-GCN
@inproceedings{yan2018spatial,
title={Spatial temporal graph convolutional networks for skeleton-based action recognition},
author={Yan, Sijie and Xiong, Yuanjun and Lin, Dahua},
booktitle={Thirty-second AAAI conference on artificial intelligence},
year={2018}
}
% AAGCN
@article{shi2020skeleton,
  title={Skeleton-based action recognition with multi-stream adaptive graph convolutional networks},
  author={Shi, Lei and Zhang, Yifan and Cheng, Jian and Lu, Hanqing},
  journal={IEEE Transactions on Image Processing},
  volume={29},
  pages={9532--9545},
  year={2020},
  publisher={IEEE}
}
% MS-G3D
@inproceedings{liu2020disentangling,
title={Disentangling and unifying graph convolutions for skeleton-based action recognition},
author={Liu, Ziyu and Zhang, Hongwen and Chen, Zhenghao and Wang, Zhiyong and Ouyang, Wanli},
booktitle={Proceedings of the IEEE/CVF conference on computer vision and pattern recognition},
pages={143--152},
year={2020}
}
% CTR-GCN
@inproceedings{chen2021channel,
title={Channel-wise topology refinement graph convolution for skeleton-based action recognition},
author={Chen, Yuxin and Zhang, Ziqi and Yuan, Chunfeng and Li, Bing and Deng, Ying and Hu, Weiming},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={13359--13368},
year={2021}
}
% Shift-GCN
@inproceedings{cheng2020skeleton,
title={Skeleton-based action recognition with shift graph convolutional network},
author={Cheng, Ke and Zhang, Yifan and He, Xiangyu and Chen, Weihan and Cheng, Jian and Lu, Hanqing},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={183--192},
year={2020}
}
% DC-GCN+ADG
@inproceedings{cheng2020decoupling,
  title={Decoupling gcn with dropgraph module for skeleton-based action recognition},
  author={Cheng, Ke and Zhang, Yifan and Cao, Congqi and Shi, Lei and Cheng, Jian and Lu, Hanqing},
  booktitle={Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part XXIV 16},
  pages={536--553},
  year={2020},
  organization={Springer}
}
% MST-GCN
@inproceedings{chen2021multi,
  title={Multi-scale spatial temporal graph convolutional network for skeleton-based action recognition},
  author={Chen, Zhan and Li, Sicheng and Yang, Bing and Li, Qinghan and Liu, Hong},
  booktitle={Proceedings of the AAAI conference on artificial intelligence},
  volume={35},
  number={2},
  pages={1113--1122},
  year={2021}
}
% InfoGCN
@inproceedings{chi2022infogcn,
title={InfoGCN: Representation Learning for Human Skeleton-Based Action Recognition},
author={Chi, Hyung-gun and Ha, Myoung Hoon and Chi, Seunggeun and Lee, Sang Wan and Huang, Qixing and Ramani, Karthik},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={20186--20196},
year={2022}
}