Exploiting Negative Learning for Implicit Pseudo Label Rectification in Source-Free Domain Adaptive Semantic Segmentation

TL;DR

This paper introduces PR-SFDA, a source-free domain adaptation method for semantic segmentation that uses pseudo label rectification with confidence regularization and negative learning, achieving near state-of-the-art results without source model access.

Contribution

It proposes a novel source-free domain adaptation approach employing pseudo label rectification with confidence regularization and negative learning, eliminating the need for source model internal details.

Findings

Achieves 49.0 mIoU on GTA5 to Cityscapes benchmark.

Performs comparably to methods requiring source model access.

Effectively tolerates noisy pseudo labels during training.

Abstract

It is desirable to transfer the knowledge stored in a well-trained source model onto non-annotated target domain in the absence of source data. However, state-of-the-art methods for source free domain adaptation (SFDA) are subject to strict limits: 1) access to internal specifications of source models is a must; and 2) pseudo labels should be clean during self-training, making critical tasks relying on semantic segmentation unreliable. Aiming at these pitfalls, this study develops a domain adaptive solution to semantic segmentation with pseudo label rectification (namely \textit{PR-SFDA}), which operates in two phases: 1) \textit{Confidence-regularized unsupervised learning}: Maximum squares loss applies to regularize the target model to ensure the confidence in prediction; and 2) \textit{Noise-aware pseudo label learning}: Negative learning enables tolerance to noisy pseudo labels in training, meanwhile positive learning achieves fast convergence. Extensive experiments have been performed on domain adaptive semantic segmentation benchmark, \textit{GTA5 $\to$ Cityscapes}. Overall, \textit{PR-SFDA} achieves a performance of 49.0 mIoU, which is very close to that of the state-of-the-art counterparts. Note that the latter demand accesses to the source model's internal specifications, whereas the \textit{PR-SFDA} solution needs none as a sharp contrast.