Remote Sensing ChatGPT: Solving Remote Sensing Tasks with ChatGPT and Visual Models

• DOI: 10.48550/arxiv.2401.09083
• Publication Date: 2024-01-01
• Authors: Haonan Guo; Xin Su; Chen Wu; Bo Du; Liangpei Zhang; Deren Li

Recently, the flourishing large language models(LLM), especially ChatGPT, have shown exceptional performance in understanding, reasoning, and interaction, attracting users researchers from multiple fields domains. Although LLMs great capacity to perform human-like task accomplishment natural image, their potential handling remote sensing interpretation tasks has not yet been fully explored. Moreover, lack of automation planning hinders accessibility techniques, non-remote experts research fields. To this end, we present Remote Sensing an LLM-powered agent that utilizes ChatGPT connect various AI-based models solve complicated tasks. More specifically, given a user request utilized understand requests, according tasks' functions, execute each subtask iteratively, generate final response output subtask. Considering LLM is trained with capable directly perceiving visual concepts as contained images, designed cues inject information into ChatGPT. With can simply send image corresponding request, get results well feedback Experiments examples show tackle wide range be extended more sophisticated such foundation model. The code demo publicly available at https://github.com/HaonanGuo/Remote-Sensing-ChatGPT .