Self-Correlation and Cross-Correlation Learning for Few-Shot Remote Sensing Image Semantic Segmentation
This is the source code for our paper Self-Correlation and Cross-Correlation Learning for Few-Shot Remote Sensing Image Semantic Segmentation, which is accepted to Sigspatial 2023.
- Install PyTorch 2.0.1 and other dependenies
- Clone this repo
git clone https://github.com/linhanwang/SCCNet.git
Download remote_sensing.tar.gz from here, unzip and put it under your directory 'SCCNet'.
python train.py --max_steps 200000 --freeze True --datapath './remote_sensing/iSAID_patches' --img_size 256 --backbone resnet50 --fold 0 --benchmark isaid --lr 9e-4 --bsz 32 --logpath exp_name
The log and checkpoints are stored under directory 'logs'.
python test.py --datapath './remote_sensing/iSAID_patches' --img_size 256 --backbone resnet50 --fold 0 --benchmark isaid --bsz 64 --nshot 1 --load './logs/exp_name/best_model.pt'
We provide a pretrained-model for the above setting. You can download it from here.
The fusion process is implemented in test.py, you can turn it on in test.sh. You can turn it on by setting fuse=True. The option eigen_path is used to indicate the directory of eigen vectors. The process to generate eigen vectors is described in spectral.