This is pytorch implementation for human Reid described in the paper: Recurrent Convolutional Network for Video-based Person Re-Identification
- Pytorch Version --v0.4 with CUDA > 8.0
- Numpy --v1.14
- OpenCV --v3.2
- Matplotlib --v2.1
First we need to split the data into train and test
- Download the iLIDS-VID dataset.
- Run the following command:
python prepare_data.py --dataset_dir=/path/to/i-LIDS-VID/sequences --data_name=<dataset_name>
Note: In order to see other changeable parameters such as gen_opt_flow, train_test_split, and frames_per_step run the following command:
python prepare_data.py --h
Once the data is successfully prepared, the model can be trained by running the following command:
python train.py --dataset_dir=/path/to/i-LIDS-VID/sequences --dataset_name=<dataset_name>.txt --checkpoint_dir=/where/to/store/checkpoints
Note: In order to see other changeable parameters such as batch size, image height/width, sequence length, etc., run the following command:
python train.py --h
In order to see the training loss graph open a tensorboard
session by
tensorboard --logdir=./runs/<log_folder> --port 8080
Once model is trained, we can compute cmc by running the following command:
python rankCMC_test.py --dataset_dir=/path/to/i-LIDS-VID/sequences --checkpoint_dir=/where/checkpoints/stored --checkpoint_file=hnRiD_latest --n_steps=<number of steps>
Note: In order to see other changeable parameters such as image height/width, use_data_aug, etc, run the following command:
python rankCMC_test.py --h
Original Code https://github.com/niallmcl/Recurrent-Convolutional-Video-ReID
@inproceedings{mclaughlin2016recurrent,
title={Recurrent convolutional network for video-based person re-identification},
author={McLaughlin, Niall and del Rincon, Jesus Martinez and Miller, Paul},
booktitle={Computer Vision and Pattern Recognition (CVPR), 2016 IEEE Conference on},
pages={1325--1334},
year={2016},
organization={IEEE}
}