whuhxb / changebind

ChangeBind: A Hybrid Change Encoder for Remote Sensing Change Detection

This repo contains the official PyTorch code for ChangeBind.

Introduction

ChangeBind utilizes a change encoder that leverages local and global feature representations to capture both subtle and large change feature information to precisely estimate the change regions.

➡️ Requirements

pytorch 1.10.0
timm 0.4.12
opencv-python
tqdm
pillow

➡️ Data structure

"""
Change detection data set with pixel-level binary labels；
├─A
├─B
├─label
└─list
"""

A: images at time frame t1;

B:images at time frame t2;

label: label masks;

list: contains train.txt, val.txt and test.txt, each file records the image names (____.png) in the change detection dataset.

Citation

@inproceedings{changebind2024,
  title={ChangeBind: A Hybrid Change Encoder for Remote Sensing Change Detection},
  author={Noman, Mubashir and Fiaz, Mustansar and Cholakkal, Hisham},
  booktitle={Arxiv},
  year={2024}
}

Acknowledgements

Thanks to the codebases [ScratchFormer] [BIT] [ChangeFormer].

See Also

ScratchFormer: Remote Sensing Change Detection With Transformers Trained from Scratch

ELGCNet: Efficient Local-Global Context Aggregation for Remote Sensing Change Detection