Mainly:
- Anaconda
- Pytorch
- OpenCV
To run the program, you should install all the libraries in the file ariad.yml as follows:
conda env create -f ariad.yml
conda activate ariad
# you can run the code
When you launch the code, the dataset is partitionned between test and train through the file 'datacreator.py'. Your files must be in the directory 'Datasets'.
By default, you'll run Resnet18 model with 20 epochs
python3 train.py -backbone='resnet18' -init_lr=0.0001 -step_size=5 -nums_epochs=20 -phase='train'
To test your model, you can specify the backbone
python3 train.py -backbone='resnet18' phase='test'
python3 train.py -backbone='resnet' -phase='demo' -video='the_office.mp4' -list_imgs='mickael_scott.jpg,dwight.jpg'