A code base for computer vision in pytorch, this code is based on https://github.com/facebookresearch/SlowFast , after read the code of slowfast network, i can implement the code of computer vision and machine learning quickly. if you are new to machine learning, i wish this code base can help you.
1 before the train we should load the dataset and model, so you can first write your dataset code and model code in net/models and net/dataset and do the unit test.
2 write the hyper-parameter in net/config/defaults and the configs/xxx.yaml, you should check the loss function,optimizer and lr-scheduler.
3 take the train.py to write your code and train your model with your dataset.