Skip to content

Latest commit

 

History

History
60 lines (48 loc) · 2.27 KB

2d_hand_gesture.md

File metadata and controls

60 lines (48 loc) · 2.27 KB
Raw

2D Hand Keypoint Datasets

It is recommended to symlink the dataset root to $MMPOSE/data. If your folder structure is different, you may need to change the corresponding paths in config files.

MMPose supported datasets:

NVGesture

OneHand10K (CVPR'2016) 
@inproceedings{molchanov2016online,
  title={Online detection and classification of dynamic hand gestures with recurrent 3d convolutional neural network},
  author={Molchanov, Pavlo and Yang, Xiaodong and Gupta, Shalini and Kim, Kihwan and Tyree, Stephen and Kautz, Jan},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4207--4215},
  year={2016}
}

For NVGesture data and annotation, please download from NVGesture Dataset. Extract them under {MMPose}/data, and make them look like this:

mmpose
├── mmpose
├── docs
├── tests
├── tools
├── configs
`── data
    │── nvgesture
        |── annotations
        |   |── nvgesture_train_correct_cvpr2016_v2.lst
        |   |── nvgesture_test_correct_cvpr2016_v2.lst
        |       ...
        `── Video_data
            |── class_01
            |   |── subject1_r0
            |   |   |── sk_color.avi
            |   |   |── sk_depth.avi
            |   |       ...
            |   |── subject1_r1
            |   |── subject2_r0
            |       ...
            |── class_02
            |── class_03
                ...

The hand bounding box is computed by the hand detection model described in det model zoo. The detected bounding box can be downloaded from GoogleDrive. It is recommended to place it at data/nvgesture/annotations/bboxes.json.