Self-supervised Augmentation Consistency for Adapting Semantic Segmentation

TL;DR

This paper introduces a practical domain adaptation method for semantic segmentation that uses simple data augmentation and consistency training, avoiding complex adversarial techniques, leading to state-of-the-art results.

Contribution

It presents a lightweight, self-supervised framework that improves segmentation accuracy by enforcing prediction consistency across augmentations without adversarial training.

Findings

Achieves significant improvements over previous methods.

Consistent performance across different architectures.

Effective without additional training rounds.

Abstract

We propose an approach to domain adaptation for semantic segmentation that is both practical and highly accurate. In contrast to previous work, we abandon the use of computationally involved adversarial objectives, network ensembles and style transfer. Instead, we employ standard data augmentation techniques $-$ photometric noise, flipping and scaling $-$ and ensure consistency of the semantic predictions across these image transformations. We develop this principle in a lightweight self-supervised framework trained on co-evolving pseudo labels without the need for cumbersome extra training rounds. Simple in training from a practitioner's standpoint, our approach is remarkably effective. We achieve significant improvements of the state-of-the-art segmentation accuracy after adaptation, consistent both across different choices of the backbone architecture and adaptation scenarios.