S3Net: Innovating Stereo Matching and Semantic Segmentation with a Single-Branch Semantic Stereo Network in Satellite Epipolar Imagery

TL;DR

S3Net is a novel single-branch network that jointly improves stereo matching and semantic segmentation in satellite imagery by leveraging their intrinsic link, resulting in higher accuracy in both tasks.

Contribution

The paper introduces S3Net, an innovative multitask learning framework that integrates semantic segmentation and stereo matching in a single network using Self-Fuse and Mutual-Fuse modules.

Findings

Semantic segmentation mIoU increased from 61.38 to 67.39.

Disparity D1-Error reduced from 10.051 to 9.579.

EPE decreased from 1.439 to 1.403.

Abstract

Stereo matching and semantic segmentation are significant tasks in binocular satellite 3D reconstruction. However, previous studies primarily view these as independent parallel tasks, lacking an integrated multitask learning framework. This work introduces a solution, the Single-branch Semantic Stereo Network (S3Net), which innovatively combines semantic segmentation and stereo matching using Self-Fuse and Mutual-Fuse modules. Unlike preceding methods that utilize semantic or disparity information independently, our method dentifies and leverages the intrinsic link between these two tasks, leading to a more accurate understanding of semantic information and disparity estimation. Comparative testing on the US3D dataset proves the effectiveness of our S3Net. Our model improves the mIoU in semantic segmentation from 61.38 to 67.39, and reduces the D1-Error and average endpoint error (EPE) in disparity estimation from 10.051 to 9.579 and 1.439 to 1.403 respectively, surpassing existing competitive methods. Our codes are available at:https://github.com/CVEO/S3Net.