jshilong/GPT4RoI
A vision-language model that enables large language models to understand and reason about spatial regions within images through instruction tuning.
Not currently ranked — collecting fresh signals.
star history
GPT4RoI instruction-tunes LLaMA to process region-of-interest inputs alongside natural language instructions, enabling spatial visual understanding. The model accepts bounding box coordinates and cropped image features as input, allowing users to query specific regions with natural language. Released weights combine delta weights with original LLaMA for the 7B variant, and the project includes training code, inference code, and a Gradio demo.
Frequently asked
- What is jshilong/GPT4RoI?
- A vision-language model that enables large language models to understand and reason about spatial regions within images through instruction tuning.
- Is GPT4RoI open source?
- Yes — jshilong/GPT4RoI is an open-source project tracked on heatdrop.
- What language is GPT4RoI written in?
- jshilong/GPT4RoI is primarily written in Python.
- How popular is GPT4RoI?
- jshilong/GPT4RoI has 556 stars on GitHub.
- Where can I find GPT4RoI?
- jshilong/GPT4RoI is on GitHub at https://github.com/jshilong/GPT4RoI.