Pytorch codes of 'Cross-Difference Semantic Consistency Network for Semantic Change Detection' [paper]

Data preparation:

1. Split the SCD data into training, validation and testing (if available) set and organize them as follows:

YOUR_DATA_DIR

• Train
  • im1
  • im2
  • label1
  • label2
• Val
  • im1
  • im2
  • label1
  • label2
• Test
  • im1
  • im2
  • label1
  • label2

2. Find -datasets -RS_ST.py, set the data root in Line 22 as YOUR_DATA_DIR

3. The pretrained weights and JL1 dataset can be accessed at “Link: https://pan.baidu.com/s/1R0OZz3y2kIWCRoU_H1bQMg
   Extract code: nwpu”.

Reference

If you find our work useful or interesting, please consider to cite:

@ARTICLE{10494733,
author={Wang, Qi and Jing, Wei and Chi, Kaichen and Yuan, Yuan},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={Cross-Difference Semantic Consistency Network for Semantic Change Detection},
year={2024},
volume={62},
number={},
pages={1-12},
keywords={Semantics;Feature extraction;Task analysis;Self-supervised learning;Data models;Data mining;Solid modeling;Cross-difference;deep learning;remote sensing image;semantic change detection (SCD);semantic consistency},
doi={10.1109/TGRS.2024.3386334}}