Robust Online Aero-Engine Blade Defect Detection via Dual-Alignment Test-Time Adaptation
ABDD introduces an online adaptive detection framework for aero-engine blade defects that handles domain shifts at test time. It uses a Dual-Alignment Strategy to jointly adapt local defect morphology and global visual style. This method improves detection robustness across varying conditions like illumination and background, leading to significant performance gains in industrial inspection settings.