RS-Agent: An LLM-driven Remote Sensing Intelligent Agent

An increasing number of models have achieved great performance in remote sensing tasks with the recent development of Large Language Models (LLMs) and Visual Language Models (VLMs). However, these models are constrained to basic vision and language instruction-tuning tasks, facing challenges in complex remote sensing applications. Additionally, these models lack specialized expertise in professional domains.

To address these limitations, we propose an LLM-driven remote sensing intelligent agent named RS-Agent.

Key Features

1. LLM-powered Center Controller: RS-Agent is powered by a large language model (LLM) that acts as its "Center Controller," enabling it to understand and respond to various problems intelligently.

2. Integration of High-performance Tools: RS-Agent integrates many high-performance remote sensing image processing tools, facilitating multi-tool and multi-turn conversations.

3. Professional Knowledge Base: RS-Agent can answer professional questions by leveraging robust knowledge documents.

Experimental Results

We conducted experiments using several datasets, including:

• RSSDIVCS
• RSVQA
• DOTAv1

The experimental results demonstrate that our RS-Agent delivers outstanding performance in many tasks, including:

• Scene Classification
• Visual Question Answering
• Object Counting

Conclusion

RS-Agent addresses the limitations of existing models by providing a robust, intelligent solution for complex remote sensing applications, backed by comprehensive integration of advanced tools and knowledge bases.

Getting Started

To get started with RS-Agent, follow these steps:

1. Clone the repository:

   git clone https://github.com/yourusername/RS-Agent.git
   cd RS-Agent

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   python main.py

Contributing

We welcome contributions to RS-Agent. Please read our CONTRIBUTING.md to learn more about how to get involved.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

We thank the developers of the datasets and tools used in our experiments.

---

For more details, visit our GitHub repository.