Dependencies"

PyTorch
sikit-image
OpenCV2
NumPy
moviepy
tqdm
matplotlib

Put training data and testing data in data/train and data/test respectively; otherwise, you have to specify it using flag -p when running train.py and test.py.

Run the following command, the weights and loss graph will be stored in checkpoints and result respectively.

$ python3 train.py

Use prepare_test.py to prepare gray frames from ground truth frames and then run test.py to get the result. Result will be stored in res folder in the same path to every cut.

$ python3 prepare_test.py
$ pythone test.py

Run components.py to perform color consistency on component of character.

$python3 components.py /path/to/image_folder/ /path/to/enhanced_folder/