/FENet

The code of 'Feature Enhancement Stereo Matching Network'

Primary LanguagePython

FENet

The code of 'Feature Enhancement Stereo Matching Network'. The full code will be published after the paper is accepted.

This repository contains the highlights proposed in the paper, such as PCE module, CME module and DW loss, and provides an improved curve-fitting disparity prediction method, which is a non-differentiable method with high sub-pixel accuracy. If you use them in your work, please cite ours.

Pretrained model

https://drive.google.com/drive/folders/1d-6ZvMcweYRBglHJFgABHuYef5PTF5Mb?usp=drive_link

Cite

@article{CHEN2023104614,
title = {Feature enhancement network for stereo matching},
journal = {Image and Vision Computing},
volume = {130},
pages = {104614},
year = {2023},
issn = {0262-8856}
}

Acknowledge

GWCNet: https://github.com/xy-guo/GwcNet