Wardrobe uses AI to extract clothing from photos

Wardrobe, an open-source JavaScript project, extracts clothing items from photos using GPT-style image models, hitting 853 GitHub stars on its July 17, 2026 release [Dev.to]. The repository provides a Node-compatible library that accepts a user-uploaded image, runs it through a vision model hosted via OpenAI’s API, and returns a JSON list of detected garments with confidence scores. This output includes category (e.g., "t-shirt", "jeans"), dominant color, and a thumbnail URL, enabling developers to build digital closets, outfit generators, or usage trackers without writing custom ML pipelines [Dev.to].

The codebase is pure JavaScript, with a single-file entry point and optional TypeScript typings, making it drop-in for web and mobile apps. Documentation includes a sample React component that renders a searchable wardrobe grid directly from the API response.

Wardrobe processes images through a third-party API, meaning personal fashion data leaves the user’s device. Startups that adopt the library must implement end-to-end encryption or on-premise model hosting to meet GDPR-style regulations, a non-trivial compliance hurdle for early-stage teams [Dev.to]. By exposing a ready-made image-to-metadata pipeline, Wardrobe eliminates the need for bespoke computer-vision engineering, allowing small teams to focus on recommendation logic, UI, or sustainability features rather than model training. The project’s MIT license encourages forks and extensions, prompting a nascent ecosystem of plugins—e.g., weather-aware outfit suggestions or AR try-on widgets.