Author(s): Jaehyun Shin; Seongwook Choi; Dong Sop Rhee
Linked Author(s):
Keywords: No Keywords
Abstract: This research utilizes a method for detecting building occupation changes near riverside areas using the Bitemporal Image Transformer as the change detection model, with LEVIR-CD database used for training and detection. The model combines CNN based feature extraction with context learning based on transformers, then generates a pixel level change mask by computing and segmenting the refined feature differences between the two times. This change detection method shows to be a usable tool for identifying building changes near riverside areas.
Year: 2026