Skip to content

Latest commit

 

History

48 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Car re-identification

Authors

Usage

Dependency installation

Project dependencies are managed by poetry. To install all dependencies run:

petry install

Note

Poetry is unable to install correct PyTorch version for you. If you want to install specific PyTorch version manually (as system package for example), you can install only other dependencies (without torch) via poetry install --without system-deps.

Training

Training scripts are in train/.

Usage:

python train.py <checkpoints_path> <model_name/checkpoint_path> [--fit/--all]

Will train model_name on VeRi dataset for at least 15 epochs with batch_size=200.

Then it will be stopped by early stopping based on mAP, min_delta=0.003 and patience=5.

Top 3 checkpoints will be saved to checkpoints_path.

By default, model is trained only on VeRi dataset. If --fit is specified, training is done on FIT dataset only and if --all, training is done on both.

Testing

Testing scripts are in train/.

Usage:

python test.py <checkpoint_path>

This will evaluate model on both VeRi and FIT datasets.

Datasets

For model training, you have to obtain and your own copy for datasets. To get more information, look into datasets folder.

About

Identifikace aut podle jejich vzhledu (bez SPZ)

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages