OpenGVLab/InternGPT
InternGPT is an open-source, pointing-language-driven visual interactive system that significantly enhances user communication with AI models like ChatGPT, improving efficiency and accuracy in complex vision-centric tasks.
Core Features
Detailed Introduction
InternGPT (iGPT) is an innovative open-source platform designed to bridge the gap between human visual intuition and AI model interaction. By enabling users to communicate with chatbots like ChatGPT through direct visual inputs such as clicking, dragging, and drawing, it dramatically improves the efficiency and precision of AI in handling complex visual scenarios. The platform integrates various cutting-edge AI models, including DragGAN for interactive image editing and ImageBind for multimodal generation, and features a fine-tuned large vision-language model, Husky, achieving near GPT-4 quality in multimodal dialogue. It serves as a versatile demo platform for showcasing diverse AI capabilities.