A Five Elements Constitution Determination Method Based on Facial AI

Abstract

The application discloses a five-element constitution determination method based on facial AI, and belongs to the technical field of traditional Chinese medicine constitution recognition and artificial intelligence. In order to solve the problems of strong subjectivity, low efficiency and difficult to be applied in large scale of traditional five-element constitution determination, the application collects facial images of users, extracts facial features by using MediaPipe 468 key points, combines a preset five-element constitution determination model, maps the facial features to the traditional Chinese medicine five-element theory, and realizes automatic and standardized constitution determination. The application can be used in health management, health care guidance and other scenes, and provides personalized five-element constitution determination results and health care suggestions for users, and improves the objectivity and convenience of constitution determination.

Description

Technical Field

[0001] This invention relates to the fields of traditional Chinese medicine constitution identification, artificial intelligence image recognition, and intelligent health terminal control technology, specifically to a method for determining constitution based on facial AI using the five elements. Background Technology

[0002] With the upgrading of light wellness consumption and the modernization of traditional Chinese medicine, constitution identification based on the Five Elements theory has become an important basis for personalized wellness. Traditional Chinese medicine face diagnosis relies on the doctor's experience, which is highly subjective, lacks standardized judgment criteria, and is inefficient, making it impossible to scale up and standardize its application on smart devices. Existing constitution identification methods mostly use questionnaires, which are cumbersome, time-consuming, and result in a poor user experience; some image recognition solutions are not deeply integrated with the Five Elements theory of traditional Chinese medicine, have vague feature definitions, insufficient model accuracy, and cannot effectively link with smart beverage robots and health management devices.

[0003] Meanwhile, the practical needs for user privacy protection, local AI inference, real-time response, and device linkage are becoming increasingly prominent. Therefore, developing a standardized, contactless, rapid, privacy-secure, and smart device-linkable method for determining the Five Elements constitution has become an urgent technical problem to be solved in this field. Summary of the Invention

[0004] Purpose of the invention To address the shortcomings of existing technologies, this invention provides a method for determining the five elements constitution based on facial AI, enabling rapid, objective, standardized, and privacy-secure identification of traditional Chinese medicine constitution. It can also directly link with an intelligent herbal tea and coffee robot to complete personalized beverage recommendations and preparation, thus overcoming the deficiencies of traditional facial diagnosis and existing technologies. Technical solution

[0005] A method for determining body constitution based on facial AI includes the following steps:

[0006] The system uses the front-facing camera of a smart terminal to capture a clear, feature-complete image of the user's face under standard lighting conditions with uniform light and no strong light or backlight.

[0007] The acquired facial images are processed for face detection, alignment of key facial features, cropping, grayscale correction, noise reduction, and size normalization to unify image specifications and eliminate the influence of differences in lighting, angle, and size on the judgment results.

[0008] Extracting multidimensional key features from preprocessed images: (1) Facial features: corresponding to the five elements and five colors: green, red, yellow, white, and black; (2) Gloss characteristics: moist, dry, oily, dull, etc.; (3) Texture characteristics: characteristics such as skin tightness, roughness, and undulation; (4) Regional chromaticity characteristics: chromaticity distribution of the forehead, cheeks, nose and chin.

[0009] The extracted features are input into a lightweight convolutional neural network model, which performs inference on the local terminal without transmitting the original face image back to protect user privacy. The model outputs five constitution types (metal, wood, water, fire, earth) and simultaneously determines the state of cold / heat, dampness / dryness, and qi / blood.

[0010] Based on the user's constitution type, the system outputs a suitable herbal tea or coffee beverage and generates control instructions such as extraction temperature, time, and concentration, which are then sent to the intelligent beverage robot for preparation. At the same time, it establishes a user constitution profile, records judgment data and drinking records, and supports cloud synchronization and cross-device recognition.

[0011] Wood type constitution: bluish complexion, easily irritable, stressed, and easily fatigued → Soothe the liver and regulate qi, replenish qi and nourish the body's vital energy; Fire-type constitution: reddish complexion, prone to internal heat, dry mouth, acne → Clear heat and reduce internal heat, soothe and relieve symptoms; Earth type constitution: yellowish complexion, oily skin, abdominal bloating, and excessive dampness → Strengthen the spleen, remove dampness, relieve greasiness, and aid digestion; Metal type constitution: pale complexion, dry skin, sore throat → moisturize the lungs and promote saliva production, soothe the throat and protect the voice; Water-type constitution: dull complexion, afraid of cold, weak → warm the kidneys and strengthen the body, warm the body and nourish the vital energy. Beneficial effects

[0012] Objective standardization: AI models make unified judgments, eliminating subjective differences from human consultations; Fast and seamless: The judgment is completed in 1-3 seconds, and users do not need to fill out a questionnaire, providing a smooth experience; Privacy and security: Local AI inference, without uploading original facial images, complies with data compliance requirements; Strong equipment integration: Directly connects to intelligent tea and coffee robots to form a closed loop of "physical constitution identification - beverage customization - health management"; Highly adaptable: It can be deployed on a large scale in shopping malls, office buildings, communities, and elderly care institutions, and is applicable to a wide range of scenarios. Detailed Implementation

[0013] The robot uses a front-facing high-definition camera to capture the user's face, ensuring even lighting and no obstructions.

[0014] The image is aligned to faces, cropped, and normalized to a size of 224×224, and then denoised and brightness balanced.

[0015] Skin color features are extracted based on the RGB / HSV color space; glossiness is calculated based on luminance variance; and skin texture is extracted based on gradient features.

[0016] Input the features into the lightweight MobileNet model and output the constitution type: Wood, Fire, Earth, Metal, Water.

[0017] The screen displays a constitution report, indicating the patient's condition (hot, cold, damp, dry), recommends suitable herbal teas and coffees, and sends the extraction parameters to the robot to complete the preparation. Industrial applicability

[0018] This invention can be directly integrated into hardware such as intelligent herbal tea and coffee robots, health retail terminals, and health and wellness service equipment without requiring significant equipment modifications. It enables low-cost, standardized, and large-scale deployment, significantly improving user experience, expanding health service scenarios, and enhancing product differentiation competitiveness. It possesses extremely high industrial applicability and market promotion value. Attached Figure Description

[0019] This invention does not include mechanical structure drawings, but submits a system framework diagram. Figure 1 A flowchart of a method for determining the Five Elements constitution based on facial AI.

Claims

1. A method for determining Five Elements constitution based on facial AI, characterized in that, Includes the following steps: S1. Acquire a frontal image of the user's face using an image acquisition device under standard lighting conditions; S2. Perform face detection, key point alignment, noise reduction, and size normalization preprocessing on the facial image; S3. Extract facial color features, gloss features, texture features, and facial region color features from the preprocessed facial image; S4. Input the extracted features into the pre-trained Five Elements Constitution Classification Model, and output the five constitution types (metal, wood, water, fire, earth) and their states of cold / heat, dampness / dryness, and qi / blood. S5. Generate suitable health drinks recommendations based on the body constitution results, and output control commands to the smart beverage device.

2. The method according to claim 1, characterized in that, The facial features corresponding to each body type in step S3 are as follows: Wood-type constitution: Characterized by a bluish complexion, tight skin texture, tense expression, and easy fatigue; Fire-type constitution: reddish complexion, prone to acne, dry and dull appearance, dry mouth; Earth type constitution: yellowish complexion, oily skin, rounded features, and dampness-related characteristics; Metal type constitution: pale complexion, dry skin, fine texture, and related characteristics of the throat and lungs; Water-type constitution: characterized by a dark or dull complexion, a cold luster, and insufficient Yang energy.

3. The method according to claim 1, characterized in that, The Five Elements Constitution Classification Model in step S4 is a lightweight convolutional neural network, using MobileNet, ShuffleNet or CNN structure, and inference is completed locally on the terminal without uploading the original facial images.

4. The method according to claim 1, characterized in that, In step S5, herbal tea / coffee drinks are automatically matched based on the body constitution results: Wood-type drinks are suitable for those that soothe the liver, regulate qi, and replenish qi and nourish the body. For those with a "fire" constitution, beverages that are cooling, soothing, and calming are recommended. Earth-type drinks are suitable for those that strengthen the spleen, remove dampness, relieve greasiness, and aid digestion. The gold element is matched with drinks that nourish the lungs, promote saliva production, soothe the throat, and protect the voice. Water-based drinks are suitable for those that warm the kidneys, strengthen the body, and nourish vital energy.

5. The method according to claim 1, characterized in that, Once the assessment is completed, a user's physical condition profile is created, recording the physical condition type, assessment time, and beverage compatibility record, and supporting cloud synchronization and cross-device recognition.

6. A Five Elements Constitution Determination System Based on Facial AI, characterized in that, include: The image acquisition module is used to acquire images of the user's face. The preprocessing module is used for face detection, alignment, and normalization; The feature extraction module is used to extract features such as skin color, gloss, texture, and chromaticity. The AI ​​inference module is used to perform the Five Elements constitution classification. The output control module is used to output the physical condition results and link them with the smart beverage device; The system is used to perform the method according to any one of claims 1-5.

Images