Robust Semantic Segmentation with Superpixel-Mix

TL;DR

This paper introduces Superpixel-mix, a superpixel-based data augmentation method that enhances the reliability, robustness, and accuracy of semantic segmentation models, especially under challenging conditions.

Contribution

The paper proposes Superpixel-mix, a novel augmentation technique that preserves object boundaries and improves semi-supervised segmentation performance and reliability.

Findings

Achieves state-of-the-art semi-supervised segmentation on Cityscapes.

Reduces network uncertainty and bias under distribution shifts.

Improves robustness against adverse weather and image corruptions.

Abstract

Along with predictive performance and runtime speed, reliability is a key requirement for real-world semantic segmentation. Reliability encompasses robustness, predictive uncertainty and reduced bias. To improve reliability, we introduce Superpixel-mix, a new superpixel-based data augmentation method with teacher-student consistency training. Unlike other mixing-based augmentation techniques, mixing superpixels between images is aware of object boundaries, while yielding consistent gains in segmentation accuracy. Our proposed technique achieves state-of-the-art results in semi-supervised semantic segmentation on the Cityscapes dataset. Moreover, Superpixel-mix improves the reliability of semantic segmentation by reducing network uncertainty and bias, as confirmed by competitive results under strong distributions shift (adverse weather, image corruptions) and when facing out-of-distribution data.