This prototype explores real-time skincare product recognition using AI-powered image analysis. It identifies and classifies products by their key functions—humectant, emollient, or occlusive—and matches them to personal skin profiles, creating a smarter in-store experience.
To make skincare selection more transparent and accessible by turning shelf scanning into instant education. The goal is to bridge dermatological science and consumer experience, reducing confusion in crowded retail spaces.
View the deployed prototype here: skinsense-prototype.vercel.app
- TypeScript / Next.js
- Tailwind CSS
- Vercel (deployment)
/model— training data and classification logic/ui— prototype interface or mockups/docs— research notes and test datamain.ipynb— notebook for running recognition models
- Open the notebook or prototype interface.
- Upload or capture an image of a product or shelf.
- View predicted product classifications and suggested matches.
(This prototype is conceptual and logic-only; no backend or live data integration is included.)
- Future integration: live camera input and ingredient database
- Add user testing and classification accuracy metrics
- Designed as part of ongoing research into human-centered AI in skincare
- Contact: http://www.danux.me