WiCoNet

Pytorch codes of 'Looking Outside the Window: Wider Context Transformer for the Semantic Segmentation of High-Resolution Remote Sensing Images' [paper]

BLU dataset [download link] [Baidu Netdisk]

To be updated: - [x] Codes for the BLU dataset - [x] Codes for the GID - [x] Codes for the Potsdam dataset - [ ] Optimizing the codes to easily switch datasets

How to Use 1. Split the data into training, validation and test set and organize them as follows:

YOUR_DATA_DIR - Train - image - label - Val - image - label - Test - image - label

1. Change the training parameters in Train_WiCo_BLU.py, especially the data directory.

2. To evaluate, change also the parameters in Eval_WiCo_BLU.py, especially the data directory and the checkpoint path.

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

L. Ding et al., "Looking Outside the Window: Wide-Context Transformer for the Semantic Segmentation of High-Resolution Remote Sensing Images," in IEEE Transactions on Geoscience and Remote Sensing, doi: 10.1109/TGRS.2022.3168697.