Copyright: original works, welcome to forward! Forwarding attached link https://blog.csdn.net/qq_26369907/article/details/90300760
First, the file structure
checkpoint
data/
__init__.py
dataset .py
get_data.py
models/
__init__.py
shuffleNetV2 .py
ResNet .py
utils/
__init__.py
config .py
main .py
requirement.txt
README .md
Second, the role
1. __inti__.py role
The files in each directory must exist, if the directory with the files, directories, equivalent to a bag package.
In main.py, the application dataset.py the function loader
from data.dataset import loader
Note : If the file is added to introduce __init__.py
from .dataset import loader
In main.py, the function of the loader dataset.py
from data import loader
2. Directory effect
- checkpoint directory: store the trained model
- data directory: store data related operations, and other data preprocessing;
- models catalog: Definition Model
- utils directory: storage utility functions
- config.py file: available to all static variables stored here