Any2Any: Unified Arbitrary Modality Translation for Remote Sensing 事件

Any2Any: Unified Arbitrary Modality Translation for Remote Sensing arXiv:2603.04114v2 Announce Type: replace Abstract: Multi-modal remote sensing imagery provides complementary observations of the same geographic scene, yet such observations are frequently incomplete in practice. Existing cross-modal translation methods treat each modality pair as an independent task, resulting in quadratic complexity and limited generalization to unseen modality combinations. We formulate Any-to-Any translatio