JL1-CD: A New Benchmark for Remote Sensing Change Detection and a Robust Multi-Teacher Knowledge Distillation Framework

2025-02-19

AI Summary

Remote sensing change detection faces two major challenges: scarcity of open-source sub-meter resolution data and poor robustness in detecting multi-scale changes. To address these, we introduce JL1-CD—the first large-scale, open-source benchmark featuring sub-meter (0.5–0.75 m) registered image pairs (5,000 pairs)—and propose a Multi-Teacher Knowledge Distillation (MTKD) framework. MTKD pioneers a decoupled knowledge transfer pathway that separately models feature representation and change response, integrating remote sensing-specific registration-aware augmentation, multi-scale feature alignment, and uncertainty-aware loss. Extensive experiments demonstrate state-of-the-art performance on both JL1-CD and SYSU-CD, with consistent and significant improvements across lightweight and heavy-weight models. The code and dataset are publicly released.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Deep learning has achieved significant success in the field of remote sensing image change detection (CD), yet two major challenges remain: the scarcity of sub-meter, all-inclusive open-source CD datasets, and the difficulty of achieving consistent and satisfactory detection results across images with varying change areas. To address these issues, we introduce the JL1-CD dataset, which contains 5,000 pairs of 512 x 512 pixel images with a resolution of 0.5 to 0.75 meters. Additionally, we propose a multi-teacher knowledge distillation (MTKD) framework for CD. Experimental results on the JL1-CD and SYSU-CD datasets demonstrate that the MTKD framework significantly improves the performance of CD models with various network architectures and parameter sizes, achieving new state-of-the-art results. The code is available at https://github.com/circleLZY/MTKD-CD.

Problem

• Addresses scarcity of high-resolution remote sensing datasets
• Improves consistency in change detection across varying images
• Proposes multi-teacher knowledge distillation for enhanced performance

Innovation

• JL1-CD dataset introduction
• Multi-teacher knowledge distillation
• State-of-the-art CD performance