Official code repository for SPAct: Self-supervised Privacy Preservation for Action Recognition [CVPR-2022]
UCF101: https://www.crcv.ucf.edu/data/UCF101/UCF101.rar
HMDB51: http://serre-lab.clps.brown.edu/wp-content/uploads/2013/10/hmdb51_org.rar
VISPR: https://tribhuvanesh.github.io/vpa/
PA-HMDB51: https://github.com/TAMU-VITA/PA-HMDB51
LSHVU dataset: https://github.com/holistic-video-understanding/HVU-Dataset
cd initialization
To run initialization training for anonymization function ($f_A$):
python train_recon.py --run_id="give_any_expname_you_like"
# add --restart argument to continue the stopped training
cd anonymization_training
Load the initilization weights of $f_A$, $f_B$ and $f_T$ and start training with the following command:
python train_ssl_minimax2.py --run_id="give_any_expname_you_like"
TODO: Add code
TODO: Add code
If you find the repo useful for your research, please consider citing our paper:
@inproceedings{spact,
title={SPAct: Self-supervised Privacy Preservation for Action Recognition},
author={Dave, Ishan Rajendrakumar and Chen, Chen and Shah, Mubarak},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year={2022}
}
[1] Privacy preserving action recognition (Wu et al., TPAMI 2020): https://github.com/VITA-Group/Privacy-AdversarialLearning
[2] PA-HMDB annoatations https://github.com/VITA-Group/PA-HMDB51
[3] PyTorch Implementation of UNet: https://github.com/milesial/Pytorch-UNet
[4] Torchvision models: https://github.com/pytorch/vision/tree/main/torchvision/models