Pytorch code of CS-MCNet
Test usage: put test video in ./data/test.
python test.py test --height=720 --width=1280 --load_model_path='./checkpoints/MCNet_4_cr16/best_model.pth' --frame_num=180
default test result is in ./results and other parameters can be find in config.py.
Train usage: download our train dataset and put it in ./data/train.
python train.py train --save_train_root='./checkpoints/default_train' --cr=1/16 --weights='./weights/weights_cr16.txt' --batch_size=4 --max_epoch=50 --print_freq=10 --lr=0.01
default train result is in ./checkpoints