Skip to content

Latest commit

 

History

History
15 lines (10 loc) · 552 Bytes

File metadata and controls

15 lines (10 loc) · 552 Bytes

DynamicEarthNet Tracks - Weakly-Supervised Multi-Class Change Detection

Code used to participate in the DynamicEarthNet challenge host by Earth vision 2021 challenge with simple change detection deep learning models

Dataset could be found at : https://dynearthnet.vision.in.tum.de/

You could use either Differential siamese model or multi data Unet. All paths are customizable from the configuration file

Requirements: python 3.6+, Pytorch

Run training script to get weights : python train.py Run inference on test set : python test.py