Instructions on how to run our code

Ensure that you have the following dependecies

• pytorch >= 1.10
• torchvision
• pyyaml
• tensorboard
• opencv-python
• tqdm
• py7zr
• kornia
• Latest version of C++.

Clone OpenGait repository

$ git clone https://github.com/ShiqiYu/OpenGait.git

Ensure that OpenGait and its dependecies are installed correctly.

Dataset

Follow through OpenGait preprocessing instructions for the OU-MVLP dataset and place the .pkl file in the ./data folder.

$ mv PATH_TO_PKL_FILE ./data/OUMVLP.pkl

OpenGait modifications

Apply our modifications to OpenGait to enable gender analysis:

$ bash ./src/move_files.sh

Run the evaluation script

This script will create the score matrices required for gender analysis:

bash ./OpenGait/test.sh

Experiments

To replicate the results discussed in the paper, run the following commands:

jupyter nbconvert --execute src/split.ipynb --to html   # generate random strata

bash ./src/experiment.sh                                # compute genuine/impostor distributions

jupyter nbconvert --execute src/plot.ipynb --to html    # plot the results

This will create a file plot.html with the results in the same format as presented in the paper (histograms, ROC curves, heatmaps, and $d_B$+$EER$ values).

Citations

Please cite our work, as well as OpenGait.

@InProceedings{Ibragimov_2024_FG,
    author    = {Ibragimov, Azim and Pamplona Segundo, Mauricio and Sarkar, Sudeep and Bowyer, Kevin},
    title     = {Unveiling Gender Effects in Gait Recognition using Conditional-Matched Bootstrap Analysis},
    month     = {May},
    year      = {2024},
}

@InProceedings{Fan_2023_CVPR,
    author    = {Fan, Chao and Liang, Junhao and Shen, Chuanfu and Hou, Saihui and Huang, Yongzhen and Yu, Shiqi},
    title     = {OpenGait: Revisiting Gait Recognition Towards Better Practicality},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2023},
    pages     = {9707-9716}
}