Weakly Supervised Change Detection via Knowledge Distillation and Multiscale Sigmoid Inference

arXiv:2403.05796v1 Announce Type: new
Abstract: Change detection, which aims to detect spatial changes from a pair of multi-temporal images due to natural or man-made causes, has been widely applied in remote sensing, disaster management, urban management, etc. Most existing change detection approaches, however, are fully supervised and require labor-intensive pixel-level labels. To address this, we develop a novel weakly supervised change detection technique via Knowledge Distillation and Multiscale Sigmoid Inference (KD-MSI) that leverages image-level labels. In our approach, the Class …

abstract arxiv change cs.cv detection disaster disaster management distillation etc however images inference knowledge labor management natural sensing sigmoid spatial temporal type urban via