1 paper · 1 filter
Sungjune Park, Yeongyun Kim, Se Yeon Kim +1
Large Vision and Language Models (LVLMs) have shown strong performance across various vision-language tasks in natural image domains. However, their application to remote sensing (…