Exchange Is All You Need for Remote Sensing Change Detection

TL;DR

This paper addresses remote sensing change detection by challenging the need for explicit difference computation between bi-temporal images. It introduces SEED, a Siamese Encoder-Exchange-Decoder framework that uses a parameter-free feature-exchange mechanism—formalized as an orthogonal permutation—to fuse information without loss, achieving mutual-information and Bayes-risk invariance. Through extensive experiments on five datasets and three backbones, SEED matches or surpasses state-of-the-art methods while offering a simpler, unified encoder-decoder design; randomized exchange variants further regularize training without harming performance. A key finding is SEG2CD, converting segmentation models into competitive change detectors by inserting exchange alone, highlighting SEED's versatility for lightweight deployment and self-supervised pretraining. Overall, SEED provides an interpretable, robust paradigm showing that simple feature exchange suffices for high-performance information fusion in change detection, with practical implications for deployment and broader CD tasks.

Abstract

Remote sensing change detection fundamentally relies on the effective fusion and discrimination of bi-temporal features. Prevailing paradigms typically utilize Siamese encoders bridged by explicit difference computation modules, such as subtraction or concatenation, to identify changes. In this work, we challenge this complexity with SEED (Siamese Encoder-Exchange-Decoder), a streamlined paradigm that replaces explicit differencing with parameter-free feature exchange. By sharing weights across both Siamese encoders and decoders, SEED effectively operates as a single parameter set model. Theoretically, we formalize feature exchange as an orthogonal permutation operator and prove that, under pixel consistency, this mechanism preserves mutual information and Bayes optimal risk, whereas common arithmetic fusion methods often introduce information loss. Extensive experiments across five benchmarks, including SYSU-CD, LEVIR-CD, PX-CLCD, WaterCD, and CDD, and three backbones, namely SwinT, EfficientNet, and ResNet, demonstrate that SEED matches or surpasses state of the art methods despite its simplicity. Furthermore, we reveal that standard semantic segmentation models can be transformed into competitive change detectors solely by inserting this exchange mechanism, referred to as SEG2CD. The proposed paradigm offers a robust, unified, and interpretable framework for change detection, demonstrating that simple feature exchange is sufficient for high performance information fusion. Code and full training and evaluation protocols will be released at https://github.com/dyzy41/open-rscd.

Figures (13)

• Figure 1: Comparison of change-detection paradigms. (a) A Siamese encoder extracts bi-temporal features, which are then fused and decoded by a single branch. (b) Feature exchange is performed during encoding to enhance cross-temporal interaction, followed by fusion and single-branch decoding. (c) As in (b), but in addition to the fused branch, each exchanged branch is decoded separately (a triple-decoder). (d) Exchange without explicit fusion: each exchanged encoder branch is decoded separately.
• Figure 2: Feature exchange methods in remote sensing change detection.
• Figure 3: Siamese Encoder-Exchange-Decoder (SEED) Overall Architecture.
• Figure 4: Visualization results in SYSU-CD dataset.
• Figure 5: Visualization results in LEVIR-CD dataset.