CAINet

This project provides the code and results for Context-Aware Interaction Network for RGB-T Semantic Segmentation", IEEE TMM, 2023. [IEEE link] and [arxiv link] [Homepage]

Requirements

python 3.7 + pytorch 1.12.0

Network

Segmentation maps and performance

We provide segmentation maps on MFNet dataset and PST900 dataset [GoogleDrive] [BaiDu] (arn3)

Performace on MFNet dataset

Performace on PST900 dataset

Pre-trained model and testing

1. Download the following pre-trained model and put it under './checkpoint' [download checkpoint GoogleDrive] [BaiDu] (arn3)
2. run evaluate_*.py.

Citation

@ARTICLE{lv2023cainet,
  author={Lv, Ying and Liu, Zhi and Li, Gongyang},
  title={Context-Aware Interaction Network for RGB-T Semantic Segmentation}, 
  journal={IEEE Transactions on Multimedia}, 
  volume={},
  number={}, 
  year={2023},
  pages={1-13},
  doi={10.1109/TMM.2023.3349072}
  }