jjxxmiin / Deep_Learning_Project_Structure

How to make a Deep Learning Project Structure?

Deep Learning Project Structure

How to make a Deep Learning Project Structure?

/configs

• model configs
  • Choose the format
  • .json
  • .ini
  • .yaml
  • .py

/data

• your datasets
  • train
    • class 1
    • class 2
    • etc.
  • valid
    • class 1
    • class 2
    • etc.
  • test
    • class 1
    • class 2
    • etc.

/docs

• project documents
  • install
  • how to run?
  • api structure

/images

• save sample image
  • dataloader output image (+ transforms)
  • sample image for test
  • result

/checkpoints

• your trained model

/notebooks

• data processing notebook
• tutorial notebook

/scripts

• train scripts
• test scripts
• run scripts

/src

• /data

  • dataloader
  • data utils

• /model

  • model architecture
  • model utils

• /visualization

  • visualization tools

• loss.py and /loss

  • define custom loss

• optim.py and /optim

  • define custom optimizer

• scheduler.py and /scheduler

  • define custom scheduler

• transforms.py and /transforms

  • define custom data augmentation