Search papers, labs, and topics across Lattice.
To overcome the brittle coverage and lack of diversity in heuristic-based remote sensing simulation, KnowChange leverages pretrained vision-language models to infer contextually plausible spatial locations and class transitions directly from pre-change imagery. By replacing handcrafted transition rules with VLM-guided semantic reasoning, the framework produces physically coherent, diverse bitemporal training pairs within a unified generative pipeline. The resulting compact synthetic datasets consistently outperform prior synthetic benchmarks across both synthetic-to-real transfer and downstream data augmentation tasks.
Off-the-shelf VLMs can replace fragile heuristic rules for bitemporal change simulation, unlocking context-aware geospatial reasoning to yield higher-utility synthetic data at a fraction of the sample scale.
Change data synthesis provides a cost-effective solution for expanding training data and improving the performance of change detection models. However, existing synthesis methods typically rely on handcrafted rules to simulate changes, where limited coverage of class transitions restricts the diversity of synthesized data, while predefined transition designs limit their flexibility in accommodating varied change types. In this work, we introduce KnowChange, a knowledge-guided change data synthesis framework that leverages pretrained vision-language models as knowledge sources to reason about plausible change locations and class transitions from pre-change scenes and desired change types. By integrating knowledge-guided change simulation with generalizable synthesis models, KnowChange enables flexible synthesis of diverse change types within a unified framework. Extensive experiments demonstrate that KnowChange-generated data consistently outperforms existing synthetic datasets in both synthetic-to-real transfer and synthetic data augmentation, despite being generated at a compact scale. Further analyses show that the knowledge-guided change simulation can be seamlessly integrated into existing synthesis pipelines and enhance the downstream utility of synthesized data.
