FootprintNet: State-Transition-Guided Dynamic Footprint Learning for Multi-temporal Remote Sensing Change Detection

2026-07-30

AI Summary

This work addresses the limitation of existing remote sensing change detection methods, which typically assume a single change event per location and thus fail to capture the complex, recurrent dynamics of urban building changes. To this end, we introduce the Urban Building Dynamics Detection task, which models building changes through dynamic footprints—temporal intervals representing when changes occur—and provides a unified representation for both single and multiple change events. We propose FootprintNet, a novel architecture incorporating a state-action interaction mechanism and state transition constraints, leveraging temporal boundary cues to enable pixel-level dynamic footprint classification. Furthermore, we present a new evaluation metric, the Building Change Dynamics Score (BCDS), which jointly assesses semantic correctness and temporal alignment. Experiments on the TSCD, MUDS, and WUSU datasets demonstrate that our method significantly outperforms existing approaches in multi-temporal building change detection.

Abstract

Despite substantial progress in remote sensing multi-temporal change detection (MTCD), most existing MTCD methods still represent the dynamic process at each spatial location over the entire observation period using a single change category associated with the final observation. This implicit single-change assumption limits their ability to characterize regions of recurrent change closely related to human activities. To address this limitation, we introduce Urban Building Dynamics Detection (UBDD), which identifies building-change dynamic footprints, i.e., the temporal intervals in which changes occur, from multi-temporal imagery and produces pixel-wise classification masks. For regions undergoing two or more changes, UBDD introduces an independent multi-change class for unified representation, thereby enabling unified modeling of single- and multi-change processes. Furthermore, we propose FootprintNet, which abstracts building-change processes as interactions between latent states and actions, and imposes state-action transition constraints to guide the learning of causally coherent change trajectories. It further exploits temporal change-boundary cues to enhance feature contrast across boundary sides, thereby improving the discrimination among different dynamic footprints and enabling accurate detection of dynamic footprints. Moreover, we introduce the Building Change Dynamics Score (BCDS) to address the inability of conventional metrics to reflect the temporal proximity between predicted footprints and labels. It evaluates predictions according to their preservation of change semantics and temporal offsets from the corresponding labels. Extensive experiments on TSCD, MUDS, and WUSU demonstrate that FootprintNet outperforms current state-of-the-art methods. The code is available at https://github.com/zmoka-zht/FootprintNet.

Problem

• multi-temporal change detection
• recurrent change
• dynamic footprint
• building change dynamics
• remote sensing

Innovation

• dynamic footprint learning
• state-transition modeling
• multi-change detection
• temporal boundary cues
• change dynamics scoring