- Python 2.7
- Tensorflow-1.2
- The implementation files of the variants of ConvLSTM are in the local dir "patchs". You need merge them with the corresponding files of TF-1.2.
The trained models can be obtained from the below link:
Link: https://pan.baidu.com/s/1O-U_Q-5i9wxOA0MDyi3Idg Code: immi
- Convert each video files into images.
- Replace the path "/ssd/dataset" in the files under "dataset_splits"
- Use training_*.py to train the networks for different datasets and different modalities.
- Use testing_*.py to evaluate the trained networks on the valid or test subsets of Jester or IsoGD.
Please cite the following paper if you feel this repository useful.
http://papers.nips.cc/paper/7465-attention-in-convolutional-lstm-for-gesture-recognition
http://openaccess.thecvf.com/content_ICCV_2017_workshops/w44/html/Zhang_Learning_Spatiotemporal_Features_ICCV_2017_paper.html
http://ieeexplore.ieee.org/abstract/document/7880648/
@article{ZhuNIPS2018,
title={Attention in Convolutional LSTM for Gesture Recognition},
author={Liang Zhang and Guangming Zhu and Lin Mei and Peiyi Shen and Syed Afaq Shah and Mohammed Bennamoun},
journal={NIPS},
year={2018}
}
@article{ZhuICCV2017,
title={Learning Spatiotemporal Features using 3DCNN and Convolutional LSTM for Gesture Recognition},
author={Liang Zhang and Guangming Zhu and Peiyi Shen and Juan Song and Syed Afaq Shah and Mohammed Bennamoun},
journal={ICCV},
year={2017}
}
@article{Zhu2017MultimodalGR,
title={Multimodal Gesture Recognition Using 3-D Convolution and Convolutional LSTM},
author={Guangming Zhu and Liang Zhang and Peiyi Shen and Juan Song},
journal={IEEE Access},
year={2017},
volume={5},
pages={4517-4524}
}
For any question, please contact
gmzhu@xidian.edu.cn