theodorhusefest / ABD-Net

[ICCV 2019] "ABD-Net: Attentive but Diverse Person Re-Identification" https://arxiv.org/abs/1908.01114

ABD-Net adapted for Deep Learning Autonomous Vehicles

Forked repo from ABD-Net.
For details and original repo look here.

The following instructions is meant for running on EPFL clusters.

Installation

The first 2 steps are already done in the folder "scratch/husefest/".

1. Clone this repo
2. Create new virtualenv with python3 -m venv <env-name>
3. Create a folder data containing the msmst17-dataset
4. Create a folder checkpoints for training
5. Activate virtualenv with source <env-name>/bin/activate
6. Run pip install -r requirements.txt to install packages
7. Go into cd torchreid/eval_cylib and run make

Training

1. Make sure you have activated virtualenv
2. Create new screen by running screen from terminal
3. Run bash train_on_msmt.shfrom terminal

Evaluating

1. Make sure you have activated virtualenv
2. Create new screen by running screen from terminal
3. Run bash evaluate_on_msmt.shfrom terminal