04:00
2026-10-12
arxiv.org
artificial-intelligence
TerraAgent: Demonstration-Grounded Multi-Agent Reasoning for Long-Context Remote Sensing Visual Question Answering
A training-free framework called AVA, presented in arXiv paper 2610.12517v1, wraps a frozen remote sensing multimodal large language model with three modules to improve remote sensing visual question …