SkinProof

• 

  Did it work, or was it the weather?

• 

  My skin: private, export or delete anytime

• 

  Sample 2: 100% of the better oil control was the weather

• 

  Sample 1: it's working, rebuy the same active cheaper

• 

  Camera Kit capture, real weather, noise floor

• 

  YouCam AI Skin Analysis: 8 concern scores + masks

SkinProof is a product-efficacy tracker built on YouCam AI Skin Analysis. Users scan their face with YouCam Camera Kit (skincare mode); the YouCam AI Skin Analysis API returns concern scores and masks (acne, oiliness, etc.). SkinProof measures the scan's own noise floor, treats weather, humidity, PM2.5 and UV (Open-Meteo) as confounders, and gives a per-product verdict: working, not proven, or no effect, plus India-available switch suggestions with INR prices from a dated catalogue. Sample weeks are simulated and labelled as such; the scan results shown are real YouCam API outputs on an AI-generated face. Built with Vercel functions, vanilla JS, MIT licensed. Try it: https://skinproof-india.vercel.app (tap the sample toggles, then My skin).

New: zone-by-zone reading of YouCam's detection masks. Each concern's mask is split into forehead, nose, cheeks and chin relative to the face's own bounding box, so a weekly scan can say where something changed, not just by how much. On two real photos of one face in different light, the zone shares of five concerns overlap by 92% on average (an evenly spread mask would score 72%): oil control 98%, texture 96%, spots 91%, acne 90%, pores 83%. Acne favours the cheeks in both photos (58% and 55% vs 40% of the skin area). See the 'Where on the face?' card in the live app: https://skinproof-india.vercel.app

Where on the face? — SkinProof reads the YouCam AI Skin Analysis masks zone by zone (forehead, nose, cheeks, chin) for two real photos of one face; zone shares agree 92% on average vs 72% for an evenly spread mask, so the tracker can say where a concern changed.

Built With

• javascript
• open-meteo
• vercel
• youcam