ChangeFlow -- Latent Rectified Flow for Change Detection in Remote Sensing

2026-05-14

AI Summary

Existing remote sensing change detection methods struggle to model global consistency in changed regions and inadequately handle contextual dependencies and ambiguities in annotations. This work proposes a novel approach by introducing latent-space rectified flows to reformulate the task as a generative modeling problem. A lightweight, structured conditional signal guides the rectified flow to synthesize change masks, enabling multi-sample prediction for uncertainty estimation. The method achieves significant improvements in global consistency and robustness while maintaining fast inference. Evaluated on four benchmark datasets, it attains an average F1 score of 80.4%, outperforming the current state-of-the-art by 1.3 percentage points.

Abstract

Remote sensing change detection (RSCD) aims to localise changes between two images of the same geographic region. In practice, change masks often follow region-level annotation conventions rather than purely local appearance differences, making them context-dependent and occasionally ambiguous. Most state-of-the-art methods utilise per-pixel discriminative classification, which produces a single prediction per input and fails to explicitly model the changed region as a coherent whole. A natural alternative is generative formulation, which can model a distribution of plausible masks, enabling sampling to capture ambiguity and encourage global consistency. However, existing generative RSCD approaches typically lag behind strong discriminative baselines due to the high computational cost of pixel-space generation and the complexity of their conditioning mechanisms. To address the limitations of prior discriminative and generative methods, we propose ChangeFlow, a generative framework that reformulates change detection as the synthesis of a change mask in latent space via rectified flow. ChangeFlow is guided by a structured yet lightweight conditioning signal, and its stochastic design naturally supports sampling-based prediction ensembling. Namely, aggregating multiple predicted change masks improves robustness, while sample agreement provides a practical confidence estimation that highlights ambiguous regions. Across four benchmarks, ChangeFlow achieves an average F1 of 80.4\%, improving by 1.3 points on average over the previous best method, while maintaining inference speed comparable to recent strong baselines. Project page: https://blaz-r.github.io/changeflow_cd

Problem

• change detection
• remote sensing
• generative modeling
• latent space
• ambiguous annotations

Innovation

• rectified flow
• latent space
• generative change detection
• prediction ensembling
• confidence estimation