AI skin analysis should work on every skin tone, but many tools were built and tested mostly on lighter skin. The result can be redness that goes undetected, dark spots misread, or scores that simply do not fit. Here is why that happens, what matters most for brown and Black skin, and how to judge an app.
Why AI can be less accurate on dark skin
- Training data: medical and cosmetic image datasets have historically included far more light skin than dark skin, so models learn lighter skin better.
- Redness looks different: inflammation often appears purple, brown or dark red on deeper skin rather than pink, so tools tuned to pink can miss it.
- Lighting and white balance: warm or dim light shifts skin tone in photos, and some cameras handle deeper tones less well.
What matters most for melanin-rich skin
- Post-inflammatory hyperpigmentation: dark marks after spots or irritation are more common and longer-lasting on deeper skin.
- Melasma: patches of darker pigment, often on the cheeks and forehead.
- Keloids and raised scars: more likely after injury or acne in some people with darker skin.
- Sunscreen white cast: mineral filters can leave a grey cast, so the right formula matters.
How to tell if an app reads your skin fairly
- It describes skin tone with a recognised method rather than a vague label.
- It corrects for warm or cool lighting before judging colour.
- It tracks your own changes over time instead of comparing you to an average face.
- It says “not enough data” when a scan is unreliable rather than guessing.
- It is honest that it is not a medical device.
Rosee measures skin tone on your iPhone using the ITA° method, a colour measurement used in dermatology research, and corrects for lighting colour before reading redness, tone or undertone. Like any phone tool it has limits, so it compares you with your own previous scans rather than with anyone else.
Rosee scans your face on your iPhone and tracks redness, texture, tone, hydration, glow and dark circles from scan to scan, so you can see whether a change is actually helping. The photo never leaves your phone, and when the lighting is too poor to read it says “not enough data” instead of guessing. Every feature is open for your first week, then it is a monthly or yearly subscription.
Frequently asked questions
Does AI skin analysis work on dark skin?
It can, but many tools were trained mostly on lighter skin and can miss redness or misread pigment on brown and Black skin. Look for apps that measure tone properly, correct for lighting and track your own changes over time.
Why do skin apps miss redness on darker skin?
Inflammation often looks purple, brown or dark red on deeper skin rather than pink, and tools tuned to pink tones can overlook it.
What skin concerns are more common on Black and brown skin?
Post-inflammatory hyperpigmentation, melasma, keloids and raised scars are more common, and dark marks tend to last longer.
What is the ITA skin tone method?
ITA° (individual typology angle) is a colour measurement used in dermatology research to classify skin tone from light to dark based on measured colour values.