Accuracy & Fairness

Built for every shade.

Most skin AI was trained on skin it never had to serve. Oyster was built the other way round, on deeper skin tones first, and held to that standard everywhere it runs.

Cover of Oyster's computer vision and clinical skin analysis report
Technical & clinical report

Pushing the Boundaries of Computer Vision in Skin Analysis

A Technical & Clinical Analysis of the Oyster Skin Intelligence Engine.

How Oyster analyses deeper skin tones, checks photo quality and matches products. The report also explains its AI model and data practices.

Authors: Jude Jr. Chikezie, Dr Chelsea Emeh MD, Emeka Adirije and Barakat Salam.

Oyster Skin Research Lab Inc · 30 pages · PDF · 9.8 MB

The dataset

Millions of real skin images across every skin tone, weighted toward the tones the industry has historically under served.

The evaluation

Skin tone read on the 10-point Monk Skin Tone Scale, so every result can be looked at tone by tone, from the lightest skin to the deepest.

Clinical validation

Reviewed alongside dermatologists and skin therapists, so the outputs hold up in a treatment room, not just a benchmark.

Claims substantiation

When a partner makes a claim off Oyster, we give them the evidence to stand behind it. Every result is traceable.