If the code or ReSGait dataset help your research, please cite :)
@inproceedings{resgait,
title={{ReSGait}: The Real-Scene Gait Dataset},
author={Zihao Mu and Francisco M. Castro and Manuel J. Mar\'in-Jim\'enez and Nicol\'as Guil and Yan-ran Li and Shiqi Yu},
booktitle={International Joint Conference on Biometrics (IJCB 2021)},
year={2021}
}
This repo is made for benchmark experiments of our paper ReSGait: The real scene gait dataset
. This repo refers to [1] and [2]
Dataset can be found at THIS LINK
After downloading the dataset, mv the label.csv
to this project ./data
folder.
And modify the label on the first line to make it easier to use.
from:
VideoID,cloth,phoneUse,gender,carry,walkingRoute,SubjectID,ShootingDate
TO:
ID,cloth,activity,gender,carry,path,id,date
And there is a given example at "./data/label_example.csv".
First step: download and prepare the data set and labels as described pre.
Second step: change ./configs/config_example.yml
and set your own training parameter.
Third step: set train.sh
to your own config.yml
Final step: set your own testing parameter and run the test.sh
.
Benchmark experiments based on GaitSet will be open source as soon as possible. It will be available at https://github.com/zihaomu/GaitSet (Not finished.).
[1] https://github.com/pudae/kaggle-hpa (BSD License.)
[2] https://github.com/AbnerHqC/GaitSet (non-commercial License.)