← Back to Blog

AI Skin Analysis · Skin Phototypes & Privacy · 29 October 2026

Safe AI Skin Analysis for Different Skin Phototypes: Accuracy Across Skin Tones and Protection of User Images

Safe AI skin analysis for different skin phototypes rests on two pillars: reading the skin consistently on every skin tone, and protecting the facial images that users share. Dermaself is an AI skin analysis platform built around both, with color normalization developed with Politecnico di Milano, published validation results and privacy technology developed with Politecnico di Torino.

For beauty e-commerce managers and skincare retailers, both pillars build trust: shoppers follow a recommendation when the analysis reflects their own skin, and they share a selfie when they know it is handled with care.

AI skin analysis on a smartphone selfie, with skin parameters measured across different skin tones

1. Why Do Skin Phototypes and Skin Tones Matter in AI Skin Analysis?

Skin phototypes and tones matter because the same skin characteristic, such as redness or a dark spot, looks different in a photo depending on the colour of the skin, the light and the camera. An AI skin analysis platform built for a diverse customer base has to read these characteristics consistently on every complexion.

Two terms help to frame the topic:

Skin phototype

Skin tone

The visible colour of the skin. It is related to the phototype, and the same skin can look lighter, darker or redder under different light and cameras.

For beauty e-commerce and skincare retailers, the analysis of skin phototypes and tones is a business question as much as a technical one: an analysis that reflects every customer’s real skin is the basis of recommendations they trust and buy.

2. How Does Dermaself Read the Skin Accurately Across Different Skin Tones?

Dermaself makes every selfie comparable before analyzing it: it corrects light and colour, guides the user to a good photo and analyzes each area of the face separately. Computer vision, the branch of AI that analyzes images, then reads the skin on a consistent basis.

Color normalization

Developed with Politecnico di Milano, it corrects light and colour so the skin can be read reliably even outside a laboratory, whatever the shooting conditions.

Guided selfie and quality checks

On-screen guidance helps the user frame the face, and automatic checks on face position, blur and exposure make sure every photo is ready for analysis.

Analysis by facial zone

The AI maps the face and analyzes each area separately, so redness, spots or dryness are read exactly where they appear.

Consistent online and in store

The same analysis runs in the e-commerce widget and on in-store devices, so every customer gets consistent results across channels.

Inclusive skincare by design

Inclusive skincare for every skin type is one of Dermaself’s core values. The technology is designed so that every customer, whatever their skin tone, receives an analysis and a routine built on their own skin.

Tip for the best selfie: no glasses, hair tied back and even light help the AI read the skin at its best.

3. How Is the Accuracy of Dermaself’s AI Measured?

Dermaself measures its AI on images that were not used for training and releases new models only after a documented review. This way, the results reflect how the AI performs on new faces, like those of a brand’s real customers.

Dermaself internal validation, 2026

Results of Dermaself’s internal validation for specific AI functions, measured on images excluded from model training.

98%

redness assessments within one grade of the reference evaluation

92%

correct facial area identified for wrinkles

0.92–0.93

repeatability for repeated measurements of the same skin

Source: Dermaself internal validation, 2026. Redness: 121 faces. Wrinkles: more than 5,000 analysed wrinkles.

Before a new model goes live, it goes through planning, testing and a documented review, with consistency checks between selfies and in-store devices and comparisons on reference subjects.

4. How Does Dermaself Protect Users’ Facial Images?

Dermaself treats every selfie as personal data: images are stored in a private archive in Europe and can only be accessed through temporary, signed links. Protecting user images is built into every step of the skin test.

Private storage in Europe

Selfies are kept in a private archive in Europe, with no public access.

Temporary access only

Images can only be uploaded or read through signed links that expire within minutes.

On-device analysis in store

On in-store tablets, the analysis can run entirely on the device, without sending the photo to a server.

3D anonymization

Developed with Politecnico di Torino, it rebuilds faces in 3D to remove identity while keeping skin information, so the datasets that train the AI respect privacy. The technology is patent pending.

5. How Does Dermaself Handle Biometric Data and the GDPR?

The brand is the data controller

The consent screen in the skin test links to the brand’s own privacy policy, so each brand keeps control of its customers’ data.

Separate marketing consent

Sharing contact details with tools such as Klaviyo, Mailchimp or Brevo requires a separate, optional consent.

Data kept separate by brand

Each brand’s data is kept separate from every other brand’s, so customer information never crosses from one store to another.

GDPR requests on Shopify

The Shopify app handles the GDPR compliance webhooks that Shopify requires for customer data requests.

For brands, this means an AI skin analysis that fits into their existing privacy policy and consent flows, with clear control over who sees which data.

6. What Is an AI Skin Assessment Grounded in Dermatological Research?

An AI skin assessment grounded in dermatological research applies criteria developed by dermatology experts to the visible characteristics that the AI reads in a selfie. In Dermaself, these criteria come from Derming, a dermatological research centre with about 30 years of experience.

7. Which Questions Should You Ask When Comparing AI Skin Analysis Platforms?

When comparing AI skin analysis platforms, ask how they handle different skin tones and how they protect user images. These are the questions and Dermaself’s answers:

Question to ask

Dermaself’s answer

Question

How does it handle different skin tones and lighting?

Dermaself

Color normalization developed with Politecnico di Milano, guided selfie and automatic quality checks.

Question

Is accuracy measured on images not used for training?

Dermaself

Yes: 98% of redness assessments within one grade and 92% correct wrinkle area.

Question

Where are users’ selfies stored?

Dermaself

In a private archive in Europe, accessible only through temporary links.

Question

Who controls customer data?

Dermaself

The brand, as data controller, through its own privacy policy.

Question

Can the analysis run without sending photos to a server?

Dermaself

Yes, on in-store tablets the analysis runs on the device.

Question

Is the analysis grounded in dermatological research?

Dermaself

Yes, with criteria from Derming and more than 60 skin parameters.

Safe AI Skin Analysis for Every Skin Tone with Dermaself

Dermaself combines an AI skin analysis designed for every skin tone, with color normalization and quality checks, and advanced protection of user images, from private storage in Europe to on-device analysis in store. The AI selects the most suitable products for each customer from the brand’s catalog, online and in store, while each brand keeps full control over its customers’ data.

Martina Mazzolin, Social Media Coordinator at Dermaself

Written by

Martina Mazzolin

Social Media Coordinator, Dermaself

Martina Mazzolin is Social Media Coordinator at Dermaself, a Milan-based beautytech company, where she manages social media channels, identifies emerging trends and develops content strategies to strengthen brand awareness and community engagement. She focuses on organic social media growth, content creation and performance analysis, contributing to the development of Dermaself's digital presence and online community. She holds a Bachelor's degree in Communication Sciences and Technologies from the University of Ferrara and a Master's degree in Communication Theory and Technology from the University of Milano-Bicocca.

LinkedIn →

Frequently Asked Questions

Safe AI skin analysis across skin phototypes: quick answers for beauty e-commerce teams and retailers

Does AI skin analysis work on every skin tone?

Dermaself’s AI skin analysis is designed to read the skin consistently across different skin tones: color normalization developed with Politecnico di Milano corrects light and colour, a guided selfie with automatic quality checks ensures a usable photo, and each area of the face is analyzed separately.

How accurate is Dermaself’s AI skin analysis?

In Dermaself’s 2026 internal validation, 98% of redness assessments were within one grade of the reference evaluation, and the AI identified the correct facial area for wrinkles in 92% of cases, on images not used for training.

Are facial images biometric data under the GDPR?

A facial image becomes biometric data under the GDPR when it is processed with specific technical means that allow a person to be uniquely identified, and biometric data used for identification is a special category of personal data under Article 9. Dermaself uses selfies to read skin characteristics for cosmetic recommendations.

Where does Dermaself store users’ selfies?

Dermaself stores users’ selfies in a private archive in Europe, with no public access: images can only be uploaded or read through signed links that expire within minutes. On in-store tablets, the analysis can run entirely on the device, without sending the photo to a server.

Who controls the data collected by the skin test?

The brand that offers the skin test is the data controller of its customers’ data: Dermaself’s consent screen links to the brand’s own privacy policy, and marketing consent for tools such as Klaviyo or Mailchimp is separate and optional.

Why does lighting change AI skin analysis results?

Lighting changes AI skin analysis results because white balance, exposure and camera processing alter how skin colour, redness and spots appear in a photo, and the effect differs between skin tones. Dermaself corrects light and colour before the analysis, so selfies taken in different conditions can be compared.

Bring Dermaself’s AI Skin Analysis into your brand’s customer experience

Integrate selfie-based skin analysis and personalized skincare recommendations into your e-commerce platform or retail store, matching customers’ skin needs with the products available in your catalog.

Book a Demo →

Contact us

Tell us about your brand or project: our team will get back to you by email.

Contact us

Tell us about your brand or project: our team will get back to you by email.