QSCP: Beyond Class-Name Prompts for Query-Guided Semantic Change Parsing

2026-09-27

AI Summary

This study addresses the limitation of existing remote sensing change detection methods that rely solely on category prompts and cannot interpret natural language queries. To overcome this, we propose a query-guided semantic change parsing framework that integrates intent parsing, semantic slot filling, bidirectional visual evidence composition, and a query-conditioned decoder. Moving beyond traditional object-class localization paradigms, our approach enables explicit reasoning over source and target states, supports synonymous expressions and complex instructions, and outputs specific masks alongside bi-temporal semantic maps. Experimental results demonstrate that the proposed framework outperforms RCDNet on the SECOND dataset, significantly improving end-to-end semantic prediction accuracy. Furthermore, evaluations on the WHU-CDC dataset validate its robust cross-domain transferability.

Abstract

Traditional change detection (CD) identifies changes between bi-temporal remote sensing images, while semantic change detection (SCD) assigns predefined land-cover classes. However, mapping all changes may not meet a user's specific needs. Referring change detection (RCD) enables selective retrieval through category prompts. However, existing category-prompted RCD uses the queried category to specify the destination of a change and returns only a binary mask of the corresponding regions. Users may instead request a particular transition and paired semantic maps to understand what changed into what. Such requests require explicit source and target reasoning beyond target-class localization. To address these needs, we propose query-guided semantic change parsing (QSCP), which supports category names, synonyms, and intent-bearing sentences and returns a query-specific mask with paired temporal semantic maps. QSCP parses requests into intents and semantic slots, composes bidirectional visual evidence, and predicts both temporal states with a query-conditioned decoder. On SECOND, QSCP outperforms RCDNet on synonym, sentence, and transition queries and improves end-to-end semantic prediction over evaluated semantic baselines. WHU-CDC experiments further assess cross-dataset transfer and consistency across equivalent expressions without target-domain training. Code is available at https://github.com/qianyuancs/QSCP

Problem

• Semantic Change Detection
• Referring Change Detection
• Remote Sensing
• Change Parsing
• Query-Guided

Innovation

• Query-Guided Semantic Change Parsing
• Intent and Slot Parsing
• Bidirectional Visual Evidence
• Query-Conditioned Decoder
• Referring Change Detection