Dermaid: nine-class skin condition classifier

MobileNetV2 transfer learning, KNUST final year project, 2024. Runs entirely in your browser: the image never leaves your device.

This is not a diagnostic tool. It was trained on 1,751 images for an undergraduate project, has not been clinically validated and has not been reviewed by a dermatologist. On a deduplicated held-out set it scores 92.7% (macro F1 0.895), but it is sensitive to lighting and framing: a close-up in warm indoor light has been misclassified with high confidence. Public dermatology datasets also skew toward lighter skin tones, so it underperforms on darker skin. If you are worried about something on your skin, see a clinician.
Selected image
Loading model, about 9 MB, one time only…