fastda 0.0.9

A simple framework for unsupervised domain adaptation

FastDA

Introduction

This is a simple framework for domain adaptation training. You can use it to build your own training process. It heavily relies on MMCV since we use a lot of useful tools (e.g., Config, Registry, Hook). The main difference between FastDA and MMCV is the Runner class. MMCV provides two kinds of runners to control the training and validation process, namely EpochBasedRunner and IterBasedRunner .

Installation

1. Prepare environment: Install pytorch and mmcv .

pip3 install torch
pip3 install mmcv

Note: Since MMCV requires Python 3.6+, our FastDA also maintains this requirements.

2. Install FastDA

pip3 install fastda

License

This project is released under the MIT License.