SkinCheck

One selfie in, a plain-language skin routine check out — 12 skin scores, an overall score, and an estimated skin age from the YouCam Skin AI API.

Comment

The problem. AI skin analysis is powerful but usually buried inside brand apps and shopping flows, and its output is raw numbers that don't tell an everyday user what any of it means or what to do next. SkinCheck strips it down to one question: "How does my skin look in this photo, and what are the sensible next steps?"

What it does.

  1. Upload or capture one front-facing face photo (on mobile, the file input opens the camera).
  2. SkinCheck sends the photo to the YouCam AI Skin Analysis API and shows the results as simple 0–100 bars: 12 skin metrics (pores, dark circles, moisture, eye bags, firmness, oiliness, age spots, acne, radiance, wrinkles, texture, redness), each with a link to YouCam's detection overlay for that concern.
  3. Alongside the metrics it shows an overall skin score and an estimated skin age.
  4. The lowest-scoring areas are turned into a short "routine check" — plain-language, rule-based next steps (for example: gentle cleansing and avoiding heavy products for visible pores; sleep, hydration, and sun protection for dark circles). No diagnosis, no product pitches — just common, sensible basics matched to what the photo shows.

No accounts, no payments, no database. One page, one photo, one results screen.

How it uses the YouCam API. SkinCheck runs the full documented YouCam AI Skin Analysis V2 server-to-server flow from a single Vercel serverless function (/api/skin), which keeps the API key server-side:

  1. POST /s2s/v2.0/file/skin-analysis — initializes the file upload and returns a file_id plus a pre-signed upload URL.
  2. The photo bytes are uploaded to the pre-signed URL.
  3. POST /s2s/v2.0/task/skin-analysis — creates the analysis task with the SD action set and returns a task_id.
  4. GET /s2s/v2.0/task/skin-analysis/<task_id> — polled until the task succeeds; the browser keeps polling through the proxy if the function returns while the task is still running.

The results UI renders whatever score fields the live response actually contains (it normalizes both documented response shapes, including the all overall score and skin_age entries), rather than assuming a fixed field list.

What was built in the hackathon window (Oct 30 – Nov 2, 2026). Everything: the API research and live integration, the single-page app, the serverless proxy implementing the async file/task flow, the dynamic results rendering, and the rule-based routine-check suggestions. The integration was tested end-to-end against the live API with a real photo before the UI was finalized, so every score shown in the screenshots and demo video is a real API result.

Privacy. Your photo is sent only to the YouCam skin analysis service and is processed within your browser session; SkinCheck itself does not store your photo or your results. Per YouCam's API terms, uploaded files and results are automatically deleted after 24 hours.

Disclaimer. SkinCheck provides general information about how skin appears in a photo. It is not a medical diagnosis and is not medical advice. For any skin concern that persists or worries you, talk to a dermatologist.

Built With

Share this project:

Updates

Submission history