JL1-CC&QA: Extending the JL1-CD Benchmark with Change Captioning and Question Answering

2026-06-30

AI Summary

This work addresses the limitation of existing remote sensing change detection methods, which typically produce only pixel-level binary segmentation without semantic interpretation of changes. To bridge this gap, the authors establish the first unified multi-task benchmark for remote sensing change understanding based on the JL1-CD dataset, introducing two novel tasks: change captioning (CC) and change-based visual question answering (QA). For the first time, binary masks, natural language descriptions, and interactive QA are jointly modeled on the same image set. Through a three-stage pipeline involving multimodal large model generation, vision-guided LLM evaluation, and human expert validation, the authors release a high-quality dataset comprising 17,021 change descriptions and 20,060 QA pairs spanning eight question types, enabling fine-grained semantic analysis of changes in remote sensing imagery.

Abstract

Remote sensing change detection (CD) traditionally focuses on pixel-level binary segmentation, which identifies where changes occur but neither what nor why. To bridge this semantic gap, we introduce JL1-CC&QA, a multi-task benchmark that extends the JL1-CD dataset with two complementary annotation layers: change captioning (CC) and change question answering (QA). Built upon 5,000 bi-temporal image pairs acquired by the Jilin-1 satellite at 0.5-0.75m ground sample distance, the benchmark comprises: (i) JL1-CC, providing 17,021 quality-verified captions that describe diverse land-cover transformations; and (ii) JL1-QA, offering 20,060 question-answer pairs across eight question types, enabling fine-grained, interactive interrogation of surface changes. All annotations are produced via a three-stage pipeline consisting of multi-modal large language model (LLM) generation, vision-grounded LLM judging, and human expert verification. We hope that JL1-CC&QA, as a benchmark unifying binary change masks, change captions, and change-oriented QA over the same image set, will serve as a valuable resource for the community to advance multi-task change understanding in remote sensing. The dataset is available at https://github.com/circleLZY/JL1-CD.

Problem

• change detection
• semantic gap
• remote sensing
• change captioning
• question answering

Innovation

• change captioning
• change question answering
• multi-task benchmark
• remote sensing change detection
• vision-grounded LLM