Skin Analysis API Functions

Product Overview

AI Skin Analysis, developed in collaboration with dermatology experts and professional

institutions in the skincare industry, combined with years of experience in the AI vision

industry, enables scientific and precise skin analysis on mobile devices. The analysis includes

skin dryness/oiliness, sensitivity analysis, pigmentation analysis, aging analysis, and acne

detection and analysis. The product functions are based on an API, which can be integrated

into multiple platforms (mobile SDK, web), empowering skincare product sales and medical

beauty businesses to upgrade their AI-driven digital marketing strategies.

Product Features

Providing professional analysis results across up to 26 skin attributes and issues, covering 7

key dimensions, with detailed and quantifiable data.

Product Specifications

The AI Skin Analysis API suite is divided into three versions: Basic, Advanced, and

Professional, offering different levels of services ranging from lightweight to professional

solutions.

Comparison of the functional items of the Basic, Advanced and Professional editions

Skin
Function items Skin analyze Advanced Skin analyze Pro
analyze

Double eyelid test for left

YES YES
and right eyes

YES
Acne Detection YES
Location Analysis

Pore Detection YES YES YES

 Skin
Function items Skin analyze Advanced Skin analyze Pro
analyze

Degree analysis

YES

Skin Type YES YES Analysis of the degree of oil

production

YES YES

Face Fine Lines Detection YES Analysis of the degree Analysis of the degree of

of forehead lines forehead lines

Eye bags, dark circles YES YES

YES
detection Degree, type analysis Degree, type analysis

YES YES
Mole and spot detection YES
Location Analysis Detailed location analysis

YES YES
Blackhead Detection YES
Degree analysis Extent, quantitative analysis

Skin color classification YES YES

Skin tone classification YES YES

Skin Age Analysis YES YES

YES
Closure detection YES
Acne analysis, detailed

YES
Skin Sensitive Area YES
Red zone, brown zone, texture
Detection Red Zone
enhancing pores and lines
 Skin
Function items Skin analyze Advanced Skin analyze Pro
analyze

Sore Control Testing YES

Number of fine lines and

YES
area detection

Analysis of the degree of

YES
skin pigmentation

Comparison of the inspection items of Basic, Advanced and Professional editions

Skin analyze
Category Inspection items Skin analyze Skin analyze Pro
Advanced

Skin Type IV
YES YES YES
Classification

YES

Oil-out area Oil shine detection

Skin Type detection chart, area share &

severity

YES

Moisture testing Moisture detection map,

area share & severity

Five categories of
YES YES YES
skin color
Skin color
Skin color ITA
YES YES
classification
 Skin Tone HA
YES YES
Classification

YES
YES
Pore size Size & number &
Pore size YES
detection severity classification &
detection
score

YES YES

Blackhead Detects the Size & number &

YES
Detection presence or severity classification &

Roughness absence of score

Pore blackhead

black and white

YES
enhancement

chart

YES

Texture detection Texture area percentage

& severity score

YES
YES
Pigmentation Detects the YES
Rectangular area +
Detection presence or Rectangular area
polygon
Pigmentation absence of

YES

Pigmentation Percentage of

pigmentation area &

 severity score, area

detection map, contour

coordinates

YES
YES
Detects the
Mole Detection YES Rectangular box +
presence or
polygon
absence of

Acne with or
YES
without

YES
Acne Regional YES
Rectangular area +
Detection Rectangular area
polygon

YES
Acne
YES Detection of occlusive

Detects acne and acne papules, pustules,

Acne grading xerostomia only, nodules, output

output rectangular area +

rectangular area polygon, severity and

score

Red Zone Map YES YES

YES

Sensitivity Red Zone Sensitive area

Detection coordinates + polygon

box
 Sensitive area size YES YES

Sensitivity level YES YES

YES

12 detailed parameters
Raised Head Lines YES YES
such as severity and

score

YES YES

Normal lines (left With or YES 12 detailed parameters

and right) without Severity such as severity and

classification score

YES

Crow's feet (left 12 detailed parameters

YES YES
and right) such as severity and
Senility
score

YES
Fine lines under
12 detailed parameters
the eyes (left and YES YES
such as severity and
right)
score

YES
Corners of the
12 detailed parameters
mouth lines (left
such as severity and
and right)
score

Lines between the YES

eyebrows 12 detailed parameters

 such as severity and

score

YES

12 detailed parameters
forehead wrinkles
such as severity and

score

YES

Cheek lines (left 12 detailed parameters

and right) such as severity and

score

Wrinkles & Fine YES

Lines Detection Severity and mapping

YES

Jawline Mandibular line angle

and coordinates

YES

Apple muscle Apple muscle

coordinates

YES

Dark Eye Circles Left and right eye

YES YES
Classification severity, type, and
Eye Problems
contour line coordinates

Eye bags YES YES YES

classification With or Severity Severity, contour line

 without coordinates

classification

Contour color
Customization YES
customization

Scoring Individual
YES
System function scores

Face coordinate
YES
values

Face Size Ratio YES

Percentage of
YES
Image quality bangs

Face angle value YES

Determine

whether glasses YES

are worn.

Introduction to Skin Analysis Algorithm Capabilities

The AI Skin Analysis algorithm encompasses seven major categories: dryness/oiliness,

roughness, pigmentation, acne, sensitivity, aging, and eye-related issues, with detection

capabilities covering up to 26 skin attributes and concerns.

Dryness/Oiliness Analysis:

Skin Type Classification:

Applicable Versions: Basic, Advanced, Professional

Using deep learning algorithms, the system analyzes facial images to classify skin types

based on dryness or oiliness. The API outputs the classification category and its confidence

score.

Case:

Oily Area Extraction:

Applicable Version: Professional

By extracting the oil reflection characteristics from facial images taken under flash lighting,

the system can analyze the oiliness of the T-zone and cheeks. It provides a numerical

measurement of the oily areas.

Case:
Roughness Analysis

Blackhead Severity and Enlarged Pores Detection

Applicable Versions: Basic, Advanced

Blackhead and Pore Mapping Analysis

Applicable Version: Professional

Through texture recognition processing of facial photos, the system extracts and highlights

enlarged pores on the cheeks and T-zone, as well as prominent blackheads on the nose. The

API can output a pore and blackhead detection map (texture_enhanced_pores), and provide
the number of enlarged pores, the number of blackheads, and the ratio of enlarged pores in

the T-zone, cheeks, and chin.

Pigmentation Analysis

Skin Tone Detection - East Asian Five-Class Skin Tone

Applicable Versions: Advanced, Professional

Based on common skin tones among East Asians, the system classifies skin tones into five

categories:Translucent Fair,Light, Natural, Tan, and Dark (Bronzed). This classification covers

a wide range of skin tones for both East Asian men and women and can be used in scenarios

such as foundation shade selection and makeup tone matching.

Skin Tone Detection - ITA Skin Tone Classification

Applicable Versions: Advanced, Professional

Based on the Fitzpatrick skin type classification defined by the FDA, this is an internationally

recognized standard for skin tone. It uses the measurement of color attributes in the Lab

color space to classify skin tones. The higher the ITA value, the higher the concentration of

pigment distribution in the skin, making it less prone to sunburn but more likely to develop
localized pigmentation after sun exposure. The lower the ITA value, the lower the

concentration of pigment, making the skin more prone to sunburn but less likely to develop

localized pigmentation (such as dark spots).

Through calibration under conditions using a mobile phone's rear flashlight, the API

outputs the ITA angle value along with the corresponding skin tone classification.

This method is highly dependent on ambient lighting. We recommend using

high-definition facial photos taken with flash lighting for accurate processing.

Measurements taken under natural or low-light conditions may result in inaccurate or

abnormal ITA values.

Skin Tone Detection - HA Skin Tone Hue Classification

Applicable Versions: Advanced, Professional

Similar to ITA, the HA classification outputs an angle value that indicates whether the skin

tone leans more towards red or yellow. A higher HA value indicates a more yellowish skin

tone with deeper blood vessel distribution, which makes the skin more prone to
pigmentation from glycation and oxidation. A lower HA value indicates a reddish skin tone

with more superficial blood vessel distribution and thinner keratin layers, making the skin

more prone to sensitivity and redness.

Spot and Mole Detection

Applicable Versions: Advanced, Professional

The system detects areas of facial spots and moles, outputting rectangular bounding box

coordinates for the detected spots and moles. In the Professional version, the system

additionally provides more precise polygonal boundaries for spots and moles.

Case:

(Brown circles represent spots, green circles represent moles):

Pigmentation Mapping

Applicable Version: Professional

Using the principles of dermatological imaging instruments, the system extracts brown

spots and pigmentation maps by analyzing RGB image features. This can be used in

conjunction with spot detection to mark the extent of facial pigmentation. The API outputs

the following:

 Pigmentation Detection Map (brown_area): Extracts brown spots and pigmentation

maps based on the principles of dermatological imaging instruments using RGB image

feature extraction. This map identifies the degree of facial pigmentation.。

 Pigmentation Index Detection: Based on the concentration of brown pigmentation and

the area of pigmentation on the face, the system analyzes the user's current facial

pigmentation status.

Case:
Acne Analysis

Acne and Closed Comedones Detection

Applicable Versions: Advanced, Professional

Based on a clinical dataset of 700 cases and a professionally annotated dataset of 100,000

real-life selfies, the system is further enhanced with a data augmentation set of 500,000

images. Acne features are extracted and semantically segmented through deep learning

across different color spaces. This enables accurate detection of acne locations and sizes,

with the API outputting the coordinates of acne and closed comedones within rectangular

bounding boxes.
Sensitivity Analysis

Sensitive Area Mapping and Detection

Applicable Versions: Advanced, Professional

Based on the analysis of 700 clinical sensitive skin samples and 5,000 consumer-sensitive

skin samples, combined with research on leading VISIA skin detection equipment, the

system calculates using the LAB color space. Through deep learning and image

enhancement methods, it accurately highlights the reddened areas of the skin, where the

intensity of redness represents the degree of skin sensitivity.

The API outputs the following:

 Sensitive Red Area Map (red_area): Using images captured with flash, it extracts the

inflamed, reddened areas of the face, marking them with varying shades of red to

indicate sensitivity levels. This map reflects the current inflammatory state of the skin.
 Sensitivity Analysis: Based on the redness intensity and the area of the sensitive regions,

it analyzes the user's current facial sensitivity condition.

Aging Analysis

Perceived Age (Skin Age) Detection

Applicable Versions: Advanced, Professional

Based on a dataset of 200,000 selfies from Asian men and women, combined with skin age

evaluations from medical beauty consultants and beauty advisors (BAs), the system

aggregates weighted assessments to label skin age. A deep learning model is used to

predict the perceived age of the skin. The API ultimately outputs an age value, indicating

"how old this face appears to be in the eyes of the average person."
Localized Wrinkle Analysis

Applicable Versions: Advanced, Professional

Using the same dataset of 200,000 selfies from Asian men and women, along with

evaluations from medical beauty consultants and beauty advisors (BAs) for skin age, the

system predicts skin age through a deep learning model.

The API can output detection results for several key localized wrinkles, including:

 Nasolabial Folds Severity: (Mild, Moderate, Severe)

 Forehead Wrinkles: (Present or Absent)

 Eye Fine Lines: (Present or Absent)

 Crow's Feet: (Present or Absent)

 Glabellar Lines (Frown Lines): (Present or Absent)

Case:
Fine Line Detection

Applicable Version: Professional

This feature detects fine lines on the forehead, between the eyebrows, and around the eyes.

The system highlights and maps these fine lines, providing a visual representation of the

detected lines. The API also outputs a fine line map and statistical data on the number of

lines present on the forehead and under the eyes.

Case:

Eye Area Issues

Dark Circle Classification

Applicable Versions: Professional, Advanced

The system detects and classifies dark circles under the eyes based on their color and shape.

It analyzes the type of dark circles present, providing insight into their severity and

underlying causes.

Case:
Eye Bag Classification

Applicable Versions: Advanced, Professional

The system detects the severity of eye bags and provides classification results (e.g., mild,

moderate, severe) along with confidence scores for each classification.

Case: