ChangeMamba: Remote Sensing Change Detection with Spatio-Temporal State Space Model This paper introduces ChangeMamba, a novel spatio-temporal state space model (STSS) for remote sensing change detection (CD). The proposed framework leverages the Mamba architecture, which is based on state space models, to address the limitations of traditional CNN and Transformer-based approaches in CD tasks. The Mamba architecture offers linear computational complexity and efficient processing of long sequences, making it suitable for CD tasks that require modeling spatio-temporal relationships in multi-temporal images. The paper presents three variants of ChangeMamba: MambaBCD for binary change detection (BCD), MambaSCD for semantic change detection (SCD), and MambaBDA for building damage assessment (BDA). These frameworks utilize the Visual Mamba architecture as the encoder, which allows full learning of global spatial contextual information from input images. The change decoder incorporates three spatio-temporal relationship modeling mechanisms that can be naturally combined with the Mamba architecture to achieve spatio-temporal interaction of multi-temporal features. The proposed frameworks outperform existing CNN- and Transformer-based approaches on five benchmark datasets without using complex training strategies. They demonstrate robustness to degraded data and achieve state-of-the-art performance on BCD, SCD, and BDA tasks. The source code is available for further research and development.ChangeMamba: Remote Sensing Change Detection with Spatio-Temporal State Space Model This paper introduces ChangeMamba, a novel spatio-temporal state space model (STSS) for remote sensing change detection (CD). The proposed framework leverages the Mamba architecture, which is based on state space models, to address the limitations of traditional CNN and Transformer-based approaches in CD tasks. The Mamba architecture offers linear computational complexity and efficient processing of long sequences, making it suitable for CD tasks that require modeling spatio-temporal relationships in multi-temporal images. The paper presents three variants of ChangeMamba: MambaBCD for binary change detection (BCD), MambaSCD for semantic change detection (SCD), and MambaBDA for building damage assessment (BDA). These frameworks utilize the Visual Mamba architecture as the encoder, which allows full learning of global spatial contextual information from input images. The change decoder incorporates three spatio-temporal relationship modeling mechanisms that can be naturally combined with the Mamba architecture to achieve spatio-temporal interaction of multi-temporal features. The proposed frameworks outperform existing CNN- and Transformer-based approaches on five benchmark datasets without using complex training strategies. They demonstrate robustness to degraded data and achieve state-of-the-art performance on BCD, SCD, and BDA tasks. The source code is available for further research and development.