CoRegOVCD: Consistency-Regularized Open-Vocabulary Change Detection

TL;DR

CoRegOVCD introduces a training-free framework for open-vocabulary change detection in remote sensing, improving semantic change detection accuracy across multiple benchmarks by calibrating posterior responses and enforcing spatial consistency.

Contribution

It proposes novel calibration and consistency mechanisms for training-free open-vocabulary change detection, enhancing semantic change analysis without explicit instance matching.

Findings

Outperforms previous training-free methods by 2.24 to 4.98 F1 points.

Achieves a 47.50% average F1 score on SECOND benchmark.

Demonstrates effectiveness across four diverse land-cover change detection benchmarks.

Abstract

Remote sensing change detection (CD) aims to identify where land-cover semantics change across time, but most existing methods still assume a fixed label space and therefore cannot answer arbitrary user-defined queries. Open-vocabulary change detection (OVCD) instead asks for the change mask of a queried concept. In the fully training-free setting, however, dense concept responses are difficult to compare directly across dates: appearance variation, weak cross-concept competition, and the spatial continuity of many land-cover categories often produce noisy, fragmented, and semantically unreliable change evidence. We propose Consistency-Regularized Open-Vocabulary Change Detection (CoRegOVCD), a training-free dense inference framework that reformulates concept-specific change as calibrated posterior discrepancy. Competitive Posterior Calibration (CPC) and the Semantic Posterior Delta (SPD) convert raw concept responses into competition-aware queried-concept posteriors and quantify their cross-temporal discrepancy, making semantic change evidence more comparable without explicit instance matching. Geometry-Token Consistency Gate (GeoGate) and Regional Consensus Discrepancy (RCD) further suppress unsupported responses and improve spatial coherence through geometry-aware structural verification and regional consensus. Across four benchmarks spanning building-oriented and multi-class settings, CoRegOVCD consistently improves over the strongest previous training-free baseline by 2.24 to 4.98 F1$_C$ points and reaches a six-class average of 47.50% F1$_C$ on SECOND.