SemSAM-CD: A Novel Weakly Supervised Change Detection Method Based on Semantic Guidance and Segment Anything Model Refinement

Remote sensing change detection (RSCD) provides indispensable technical support for urban dynamic monitoring. However, fully supervised methods remain limited by costly manual annotation. In such a context, weakly supervised learning (WSL) based on image-level labels has attracted increasing attenti...

תיאור מלא

מידע ביבליוגרפי

Principais autores:	Min Liu, Daifeng Peng, Yongjun Zhang, Haiyan Guan
פורמט:	Artigo
שפה:	Inglês
יצא לאור:	IEEE 2026-01-01
סדרה:	IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
נושאים:	Change detection (CD) class activation mapping (CAM) knowledge distillation (KD) segment anything model (SAM) weakly supervised learning (WSL)
גישה מקוונת:	https://ieeexplore.ieee.org/document/11302783/
תגים:	הוספת תג אין תגיות, היה/י הראשונ/ה לתייג את הרשומה!