Consistency regularization-based mutual alignment for source-free domain adaptation

Shuai Lü, Zongze Li, Xinyu Zhang, Jingyao Li

Abstract: Highlights•We propose the mutual alignment strategy.•We design the mutual alignment loss based on entropy.•We leverage information maximization and entropy to enhance predictions.•We conduct a series of experiments to demonstrate the effectiveness of CRMA.•Our method outperforms the SOTA methods on three popular benchmarks.