Develop roadmap

1. data augmentation ✔️
   • Classification ✔️
   • Segmentation ✔️
   • Detection ✔️
2. models
   • Classification: Resnet, DenseNet, ResNext ... ✔️
   • Segmentation: UNet, UNet34, UNet101, UNet152 ✔️
3. Github Action auto test ✔️
4. Dockerfile & Docker image ✔️
5. readthedocs: foucus on introduction ✔️
6. pypi package ✔️

1. training script
   • Classification
   • Segmentation
2. Simple datasat dataloader:
   • Classification: EuroSat, DeepSat
   • Segmentation: Open AI Challenge: Tanzania
3. metrics
   • Classification
   • Segmentation

1. Tensorboard
2. More Models
   • Classification:
   • Segmentation:

1. Large Dataset Loader, such as SpaceNet, xView, BigEarthNet...
2. Data Prepare tools.

1. detection models
2. detection training

1. more specific task

• change detection
• time series pixel classification(for crop type or land cover classification)

1. focus on Documents
2. bug fix
3. maybe website？

1. first stable version release

1. Multi-GPU Training?