Closet Twin is an AI-powered wardrobe assistant designed to help users manage and style their clothing through computer vision and large language models. The tool uses advanced image analysis to catalog and organize users' closets, offering personalized outfit recommendations and insights. Developed for the Build Small Hackathon community, Closet Twin aims to transform how people interact with their wardrobes by combining data-driven approaches with intuitive design. The system allows users to upload their entire closet in one click, with automatic attribute extraction for color, style, and formality. It also generates outfit suggestions based on weather, occasion, and personal preferences, helping users make informed style choices. The app's intelligent recommendation engine adapts to user feedback, ensuring continuous improvement in outfit suggestions. Closet Twin is built using open-source technologies, including the MiniCPM-V-4.6 vision model for image analysis and Gradio for the frontend interface. The tool's ability to analyze fashion inspiration images and match them to existing wardrobe items highlights its practical application in personal style management. Closet Twin represents a significant step in integrating AI into everyday life, offering a seamless and personalized experience for users looking to enhance their wardrobe organization and style choices.
The system's recommendation engine uses vision AI to extract garment attributes and models user style preferences based on interaction history. It generates outfit combinations by scoring pieces against context such as weather, occasion, and mood, while also optimizing for diversity in recommendations. Closet Twin also identifies underused items in a user's wardrobe and suggests purchases to fill gaps, helping users maximize their clothing investment. The app's interactive carousel allows users to explore multiple outfit combinations and refine choices with one-click piece swaps. Personal analytics track style preferences over time, offering data-driven insights to improve wardrobe efficiency. The platform's ability to handle incomplete data and scale from 10 to 1,000+ clothing items demonstrates its adaptability and robustness.
Closet Twin was developed as part of the Build Small Hackathon, emphasizing practical applications of multimodal AI in fashion technology. The project showcases how large language models can reason about subjective concepts like style while proving that AI can enhance rather than replace human creativity. The app's launch on HuggingFace Spaces marks its availability to the broader community, with a YouTube demo providing further insight into its functionality.
Source: huggingface
