FUSAR-GPT : A Spatiotemporal Feature-Embedded and Two-Stage Decoupled Visual Language Model for SAR Imagery

Researchers developed FUSAR-GPT, a specialized Visual Language Model for Synthetic Aperture Radar (SAR) imagery that significantly outperforms existing models. The system introduces spatiotemporal feature embedding and a two-stage training strategy, achieving over 12% improvement on remote sensing benchmarks.