AUGCO: Augmentation Consistency-guided Self-training for Source-free Domain Adaptive Semantic Segmentation

TL;DR

AUGCO is a simple, fast, and effective source-free domain adaptation method for semantic segmentation that leverages augmentation consistency and confidence to self-train on unlabeled target data.

Contribution

It introduces AUGCO, a novel self-training approach using augmentation consistency and confidence for source-free domain adaptation in semantic segmentation.

Findings

Achieves state-of-the-art results on 3 benchmarks.

Simple to implement and fast to converge.

Effective without access to source data during adaptation.

Abstract

Most modern approaches for domain adaptive semantic segmentation rely on continued access to source data during adaptation, which may be infeasible due to computational or privacy constraints. We focus on source-free domain adaptation for semantic segmentation, wherein a source model must adapt itself to a new target domain given only unlabeled target data. We propose Augmentation Consistency-guided Self-training (AUGCO), a source-free adaptation algorithm that uses the model's pixel-level predictive consistency across diverse, automatically generated views of each target image along with model confidence to identify reliable pixel predictions, and selectively self-trains on those. AUGCO achieves state-of-the-art results for source-free adaptation on 3 standard benchmarks for semantic segmentation, all within a simple to implement and fast to converge method.