How to Reduce Change Detection to Semantic Segmentation. (June 2023)

• Record Type: Journal Article
• Authors: Wang, Guo-Hua; Gao, Bin-Bin; Wang, Chengjie
• Is Part Of: Pattern recognition. Volume 138(2023)
• Publication Date: 2023-06
• Subjects: Change detection -- Semantic segmentation -- Feature fusion; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2023.109384
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: We present a new paradigm that reduces change detection to semantic segmentation which means tailoring an existing and powerful semantic segmentation network to solve change detection. Our analysis suggests that there are three possible change types within the change detection task and they should be learned separately. Therefore, we propose a MTF (Merge Temporal Features) module to learn these changes. We propose a simple but effective network, called C-3PO (Combine 3 POssible change types), detects changes in pixel-level, and can be considered as a new baseline network in this field. Abstract: Change detection (CD) aims to identify changes that occur in an image pair taken different times. Prior methods devise specific networks from scratch to predict change masks in pixel-level, and struggle with general segmentation problems. In this paper, we propose a new paradigm that reduces CD to semantic segmentation which means tailoring an existing and powerful semantic segmentation network to solve CD. This new paradigm conveniently enjoys the mainstream semantic segmentation techniques to deal with general segmentation problems in CD. Hence we can concentrate on studying how to detect changes. We propose a novel and importance insight that different change types exist in CD and they should be learned separately. Based on it, we devise a module named MTF to extract the change information and fuse temporal features. MTF enjoys high interpretability and reveals the essential characteristic of CD. And most segmentation networks can be adapted to solve the CD problems with our MTF module. Finally, we propose C-3PO, a network to detect changes at pixel-level. C-3PO achieves state-of-the-art performance without bells and whistles. It is simple but effective and can be considered as a new baseline in this field. Our code for C-3PO is available at https://github.com/DoctorKey/C-3PO .