Eye protection evaluation system for AI image analysis and evaluation method thereof

Abstract

The invention relates to the technical field of image data processing, in particular to an eye protection evaluation system and method for AI image analysis, and the system comprises an image collection module, an image processing module, an eye feature recognition module, an evaluation model analysis module and a feedback suggestion output module. The image acquisition module is used for acquiring facial image information of a user in real time through camera equipment; the image processing module preprocesses the acquired image, including image denoising, brightness adjustment, contrast enhancement, face positioning and eye region segmentation; the eye feature recognition module performs feature extraction on the processed image based on a deep learning algorithm; the evaluation model analysis module comprehensively evaluates the eye protection condition by constructing a neural network model in combination with the historical data of the user and the current image analysis result; and the feedback suggestion output module provides a personalized eye protection suggestion scheme for the user according to the evaluation result.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to an eye protection assessment system and an assessment method for AI image analysis. Background Art

[0002] With the development of the information society, electronic products (such as smartphones, tablets, and laptops) have become a widespread part of people's work, study, and daily lives, and prolonged screen time has become a normalized experience. However, issues such as prolonged close eye use, exposure to blue light from screens, and inappropriate ambient lighting conditions are leading to increasingly prominent health risks such as eye fatigue, dryness, and decreased vision, particularly among adolescents and office workers. While some eye care products and software are currently available, such as screen eye protection modes and timed break reminders, these are mostly based on static time rules or blue light filtering. They lack awareness and response to individual eye health, and are unable to provide dynamic assessment and feedback based on the user's current eye status, resulting in limited effectiveness. Furthermore, traditional vision health assessments often rely on offline examinations, which are time-consuming, costly, and limited in coverage, making them unsuitable for the growing demand for daily eye care management. Therefore, leveraging artificial intelligence (AI) to enhance eye health assessment capabilities and provide efficient and personalized eye care recommendations has become a key development direction in the field of vision protection.

[0003] In recent years, with rapid advances in image processing, deep learning, and computer vision, AI-based image recognition technology has been widely applied in fields such as medical diagnosis and behavioral analysis, demonstrating high accuracy and intelligence. In this context, eye images, as a high-value health signal source that can be collected in real time, are gaining increasing attention for their potential in early identification of conditions such as eye fatigue, dry eye syndrome, and vision loss. Some current studies have attempted to use image recognition to extract eye features (such as blink frequency, redness, and eye opening width) for fatigue detection. However, these algorithms often suffer from overly simplistic models, an inability to dynamically integrate with user behavior, and a lack of personalized adaptation for assessment models. Furthermore, existing technologies lack a closed-loop mechanism for user data collection, analysis, and feedback, hindering the development of a continuously optimized eye care intervention system. Therefore, there is an urgent need for a comprehensive AI eye care system that integrates image recognition, individual feature extraction, behavioral data fusion, and intelligent assessment recommendations. This system would not only improve assessment efficiency but also enable convenient and intelligent vision management services without the need for specialized equipment.

[0004] In view of the above situation, in order to overcome the above technical problems, the present invention designs an AI image analysis eye protection evaluation system and an evaluation method thereof to solve the above technical problems. Summary of the Invention

[0005] The technical purpose of this invention is to design an eye protection assessment system and an assessment method based on AI image analysis to improve assessment efficiency and realize convenient and intelligent vision management services without the need for professional equipment.

[0006] In order to achieve the above technical objectives, the present invention provides the following technical solutions: An AI image analysis eye protection assessment system, including an image acquisition module, an image processing module, an eye feature recognition module, an assessment model analysis module, and a feedback suggestion output module; The image acquisition module is used to collect user facial image information in real time through a camera device; The image processing module pre-processes the collected images, including image denoising, brightness adjustment, contrast enhancement, face location and eye area segmentation; The eye feature recognition module extracts features from processed images based on a deep learning algorithm; The evaluation model analysis module combines user historical data with current image analysis results to conduct a comprehensive evaluation of eye protection status by building a neural network model; The feedback and suggestion output module provides users with personalized eye protection suggestions based on the evaluation results; Among them, the image acquisition module is used to collect user facial image information in real time through a camera device, especially the image of the user's eye area, and automatically detect the acquisition angle, light intensity and distance to ensure the clarity and completeness of the image data; the image processing module pre-processes the acquired image, including image denoising, brightness adjustment, contrast enhancement, face positioning and eye area segmentation, etc., to improve image quality and subsequent analysis accuracy; the eye feature recognition module is based on a deep learning algorithm to extract features from the processed image, and identify and analyze multiple eye health parameters including eye opening frequency, blinking speed, pupil dilation, degree of red bloodshot around the eyes, eyelid lift angle, eyeball moistness, etc.; the evaluation model analysis module combines user historical data with current image analysis results, and conducts a comprehensive evaluation of eye protection status by constructing a neural network model, outputs eye protection risk level and performs trend prediction; the feedback suggestion output module provides users with personalized eye protection suggestions based on the evaluation results, combined with multi-dimensional data such as eye use time, ambient light intensity, sitting posture, etc., such as eye use time arrangement, rest frequency, eye protection exercise methods and environmental optimization suggestions. This system can be applied to young students, office workers and users who use electronic screens for a long time to enhance eye protection awareness and scientifically manage vision health. It is highly practical and intelligent.

[0007] Extract features from the collected eye images to obtain the blink frequency , pupil diameter change range , red bloodshot coverage ratio , eyelid ptosis angle , eye moistness score and other characteristic parameters; Substitute the above parameters into the eye fatigue scoring function Perform weighted summation, the formula is as follows:

[0008] in, Decibel is the weight coefficient obtained by model training, satisfying , used to balance the impact of various features on the score; The final score It is divided into multiple eye protection level intervals, forming evaluation levels such as "good", "mild fatigue", "moderate fatigue" and "severe fatigue", and corresponding eye protection recommendations are pushed according to the levels.

[0009] The image acquisition module further includes a light sensor for monitoring ambient light intensity. This sensor monitors changes in the user's ambient light intensity in real time and integrates data with the entire system. When the system detects that ambient light intensity exceeds a set threshold (e.g., too bright or too dim), the system automatically triggers a pre-set response mechanism. First, it prompts the user to adjust their eye conditions (e.g., turning on the lights or moving away from strong light sources) through voice prompts, pop-up notifications, or vibration alerts. Second, it automatically adjusts camera acquisition parameters, including but not limited to key parameters such as exposure time, gain, white balance, and contrast, to ensure that the captured images meet standard requirements for brightness, clarity, and color reproduction. This function not only effectively prevents image blur, overexposure, or underexposure caused by abnormal lighting conditions, but also significantly improves image quality stability, providing accurate and reliable raw data input for subsequent eye image recognition and AI analysis models, enhancing the scientific and accurate evaluation results. Furthermore, the module possesses a self-learning capability, optimizing parameters based on the lighting characteristics of the user's environment over time, further enhancing the intelligent and adaptable nature of image acquisition.

[0010] The image processing module possesses multi-level image optimization capabilities, utilizing a pre-trained model based on deep learning to intelligently process collected eye images. This process primarily involves image enhancement, alignment of key eye features, and deep fusion of multimodal features. In terms of image enhancement, the system intelligently repairs and optimizes image clarity, contrast, brightness, and noise, improving image quality and ensuring subsequent recognition accuracy. In terms of feature alignment, facial recognition and eye location algorithms automatically capture the eye region and geometrically correct for angular deviations and posture changes in the image, ensuring uniformity and standardization of the eye region. In the deep fusion stage, the system integrates skin color models, eyelid structure models, and age-specific eye changes to create a fusion model of features, enabling compatible processing across genders, ages, and demographics. Through this multi-level image optimization strategy, the system can adaptively handle a variety of complex data samples, avoiding recognition bias caused by individual differences. This significantly enhances the applicability, universality, and robustness of the assessment system across a wide range of populations, providing high-quality data support for subsequent eye health assessments.

[0011] The eye feature recognition module utilizes a multi-task convolutional neural network (MT-CNN) model, capable of simultaneously processing multiple eye feature tasks, enabling efficient image recognition and behavioral analysis. This module concurrently completes key tasks, including eye posture recognition, conjunctival bloodshot density detection, and blink behavior analysis, within a single model architecture. Eye posture recognition can be used to determine the user's gaze direction, eye opening width, and fatigue status; bloodshot density detection reflects the degree of eye congestion, serving as a warning signal for eye fatigue and dry eyes; and blink behavior analysis measures parameters such as blink frequency and duration to reflect the user's eye usage frequency and fatigue level. The MT-CNN model boasts powerful feature learning capabilities and real-time computing performance, enabling rapid and accurate eye feature extraction in complex backgrounds and diverse facial structures. Furthermore, the module supports automated image acquisition and analysis based on user-defined time periods (e.g., hourly or daily), continuously recording eye feature changes and forming a dynamic health profile. Through long-term tracking and trend modeling, the system can identify early signs of eye health problems in advance and provide users with scientific eye care advice and intervention measures, thus achieving truly personalized and intelligent eye health management.

[0012] The evaluation model analysis module incorporates a time series modeling mechanism to dynamically and continuously model trends in users' eye care behaviors and predict their health status. This module integrates image features collected from users at different time periods (such as blink frequency, changes in redness, and pupil constriction) with behavioral characteristics (such as daily eye use duration, frequency of electronic device use, and lighting environment records) to generate multi-dimensional time series data input. Using a recurrent neural network (RNN) as the core modeling tool, it performs deep learning and trend analysis on this time series data. RNNs are capable of capturing long-term dependencies in data, enabling them to identify potential abnormal behavioral trends or excessive eye use risks from a user's eye use patterns over time. For example, if the system detects an abnormally low blink frequency and consistently high eye use duration over multiple days, it can predict the risk of dry eye or visual fatigue, generate timely warnings, and deliver personalized eye care recommendations, such as mandatory rest reminders, lighting adjustment prompts, or eye behavior optimization solutions. This mechanism effectively improves the system's sensitivity and forward-looking judgment capabilities to changes in the user's eye protection status, realizes the transition from "passive response" to "active intervention", and significantly enhances the intelligence level of the evaluation system and user experience.

[0013] The feedback and suggestion output module, based on a knowledge graph and a rule engine, constructs an intelligent recommendation system for eye care scenarios. It can provide scientific and accurate intervention recommendations based on the user's individualized health assessment results. The system has a built-in rich eye care medical knowledge base, covering basic knowledge such as the physiological structure of the eye, common vision problems, fatigue and dry eye mechanisms, and integrates relevant national vision health standards and authoritative clinical guidance specifications to provide reliable medical support for the output of recommendations. By constructing an eye care-related knowledge graph, the system dynamically associates the user's historical image features, behavioral data, and evaluation results with knowledge nodes to achieve rapid matching of personalized recommendation paths. In terms of recommendation logic, the system integrates a rule engine to automatically trigger corresponding intervention strategies based on the user's current status (such as prolonged screen staring, abnormal ambient lighting, low blinking frequency, etc.), and outputs multimodal information forms including graphic descriptions, video demonstrations, and voice prompts to ensure the diversity and clarity of information communication. In addition, the module also supports linkage with mobile terminals or desktop devices (such as APP, web pages), automatically pushing eye protection suggestions, training tasks and behavioral reminders, improving user execution efficiency and daily compliance, and further enhancing the practicality of the system and human-computer interaction experience.

[0014] The system also includes a user identity management and data privacy protection module specifically designed to ensure comprehensive protection of user personal information, image data, and behavioral profile data throughout the assessment and feedback process. This module utilizes high-strength data encryption technologies, such as AES-256 symmetric encryption and RSA asymmetric encryption, to encrypt the storage and transmission of sensitive information, such as user-uploaded or real-time collected image data, blink rate, and eye activity records, preventing data leakage or unauthorized access over the network. Furthermore, the system supports anonymous user assessments, allowing users to use the eye care assessment service without binding their real-life identity, effectively alleviating initial privacy concerns. The module also features a fine-grained data authorization management mechanism, allowing users to customize authorization scopes, such as specifying which data can be used for model training and which data is restricted to local analysis, enhancing user control over their data. To further enhance security, the system also includes a built-in periodic data purge mechanism that automatically deletes historical data at a user-defined interval to prevent information risks from long-term accumulation. The design and operation of the entire module strictly adhere to national information security standards such as the Personal Information Protection Law and the Cybersecurity Law, as well as internationally accepted privacy protection protocols (such as GDPR), to ensure that while providing high-quality assessment services, user data privacy and system compliance and security are protected to the greatest extent possible.

[0015] The system integrates with a variety of third-party devices and platforms, enabling cross-device data collection and comprehensive analysis, thereby comprehensively improving the accuracy and practicality of eye care assessments. Specifically, the system seamlessly integrates with a variety of devices, including tablets, e-book readers, smartphones, and online learning platforms, to capture real-time data on users' eye behavior in different scenarios. This includes key information such as continuous eye use duration, screen brightness adjustment, and the type and frequency of reading content. Using this rich data input, the system can deeply analyze users' eye habits and environmental characteristics, identifying potential risk factors such as excessive eye use, inappropriate lighting, or poor reading posture. Furthermore, the cross-device collaborative monitoring mechanism enables the system to break down data silos on individual devices, enabling comprehensive tracking and dynamic assessment of users' eye behavior, improving the accuracy of the model and the scientific nature of personalized recommendations. More importantly, by integrating data from multiple platforms, the system can build a complete eye care management ecosystem, delivering eye care recommendations and reminders through multiple channels simultaneously, helping users develop healthy eye habits in different usage environments. This move not only enhances the practical value of the system, but also lays a solid foundation for creating intelligent and personalized vision health management solutions, and promotes eye protection technology towards multi-device collaboration and ecological development.

[0016] The system supports an advanced personalized learning mechanism, enabling continuous optimization based on the user's actual usage and health status. By continuously collecting user image analysis data and execution feedback, the system monitors subtle changes in the user's eye health and behavioral habits in real time, automatically adjusting the parameters of the internal model and the weighting of various evaluation indicators. This dynamic adjustment mechanism enables the evaluation system to not only adapt to individual differences among users but also evolve as their eye care behaviors and physiological conditions change. Through continuous model iteration and training, the system's prediction accuracy and adaptability are significantly improved, enabling it to more accurately reflect users' current eye care needs and potential health risks. Furthermore, this personalized learning mechanism ensures that the system effectively prevents model overfitting and performance degradation in the face of long-term data accumulation, maintaining stable and continuous evaluation capabilities. This intelligent, dynamically optimized evaluation method greatly meets the needs of personalized eye health management, providing users with scientific, personalized, and continuously effective health monitoring and guidance services, promoting the long-term maintenance and improvement of eye health.

[0017] An AI image analysis eye protection assessment method is used in conjunction with the above-mentioned AI image analysis eye protection assessment system; the method steps are as follows: Step 1: Image capture: High-definition cameras are used to capture real-time eye image information while the user is using a terminal device (such as a computer, tablet, or mobile phone). This ensures that the lighting, angle, and distance during image capture meet the set standards. Ambient light sensors are also used to simultaneously capture ambient brightness data, providing a reference for subsequent image analysis. Step 2: Image preprocessing: The raw image data is processed through operations such as denoising, color normalization, brightness compensation, and automatic eye area recognition and extraction to ensure the clarity and accuracy of image features. Image processing parameters are also personalized based on the user's background (such as skin color and glasses). Step 3: Feature extraction: Utilize a convolutional neural network model to extract deep features from eye images, including but not limited to blink frequency, pupil diameter, periocular bloodshot density, eyelid position and range of change. These features are then structured and encoded for subsequent model evaluation. Step 4: Assessment Modeling and Analysis: The extracted eye image features and historical behavior data are input into the eye protection assessment model. Based on a deep neural network and time series prediction mechanism, the model comprehensively analyzes the user's eye protection status, generates an eye protection risk level and trend chart, and provides a score for the impact of the user's current eye use behavior on vision health. Step 5. Result output and personalized recommendations: Generate personalized eye protection suggestions based on the evaluation results, and combine the user's pace of life, learning and work patterns, and environmental conditions to recommend improvement strategies such as eye time planning, rest frequency, and eye protection training methods. These suggestions are then pushed intelligently through the app, web page, or voice assistant to facilitate timely receipt and execution by users.

[0018] The beneficial effects of the present invention are as follows: (1) The present invention achieves a comprehensive and accurate assessment of the user's eye health status by introducing AI-based multi-dimensional image analysis and user behavior data fusion, significantly improving the scientific and personalized level of eye care management. The system uses a deep learning model to perform multi-level optimization and feature extraction on eye images, and can accurately identify key indicators including eye posture, bloodshot density, and blinking behavior, thereby comprehensively reflecting the user's eye fatigue level and potential health risks. Combined with time series modeling technology, the system can dynamically track the changing trends of users' eye behavior, provide early warnings of bad eye habits, and provide scientific and reasonable eye care solutions. This highly automated and intelligent eye care assessment breaks through the limitations of traditional static solutions, provides users with real-time, dynamic, and personalized health management services, and effectively improves user compliance and eye care effects.

[0019] (2) The present invention ensures the security of user personal information and health data by establishing a complete user identity management and data privacy protection mechanism, thereby enhancing the credibility of the system and user trust. The system supports the linkage of multiple devices and multiple platforms, collects rich eye environment and behavior data in real time, builds a complete eye protection ecological closed loop, and greatly expands the application scenarios and service coverage. At the same time, the personalized learning mechanism enables the evaluation model to continuously optimize itself according to user data, improving the accuracy and adaptability of long-term evaluation. This not only meets the diverse needs of different user groups, but also promotes the development of eye health management towards intelligence and dynamism, with broad application prospects and significant social value. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] The above and other aspects of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 Schematic diagram of the system structure of the present invention; Figure 2It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0022] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0023] like Figure 1-2 As shown, an AI image analysis eye protection assessment system includes an image acquisition module, an image processing module, an eye feature recognition module, an assessment model analysis module, and a feedback suggestion output module; The image acquisition module is used to collect user facial image information in real time through a camera device; The image processing module pre-processes the collected images, including image denoising, brightness adjustment, contrast enhancement, face location and eye area segmentation; The eye feature recognition module extracts features from processed images based on a deep learning algorithm; The evaluation model analysis module combines user historical data with current image analysis results to conduct a comprehensive evaluation of eye protection status by building a neural network model; The feedback and suggestion output module provides users with personalized eye protection suggestions based on the evaluation results; Among them, the image acquisition module is used to collect user facial image information in real time through a camera device, especially the image of the user's eye area, and automatically detect the acquisition angle, light intensity and distance to ensure the clarity and completeness of the image data; the image processing module pre-processes the acquired image, including image denoising, brightness adjustment, contrast enhancement, face positioning and eye area segmentation, etc., to improve image quality and subsequent analysis accuracy; the eye feature recognition module is based on a deep learning algorithm to extract features from the processed image, and identify and analyze multiple eye health parameters including eye opening frequency, blinking speed, pupil dilation, degree of red bloodshot around the eyes, eyelid lift angle, eyeball moistness, etc.; the evaluation model analysis module combines user historical data with current image analysis results, and conducts a comprehensive evaluation of eye protection status by constructing a neural network model, outputs eye protection risk level and performs trend prediction; the feedback suggestion output module provides users with personalized eye protection suggestions based on the evaluation results, combined with multi-dimensional data such as eye use time, ambient light intensity, sitting posture, etc., such as eye use time arrangement, rest frequency, eye protection exercise methods and environmental optimization suggestions. This system can be applied to young students, office workers and users who use electronic screens for a long time to enhance eye protection awareness and scientifically manage vision health. It is highly practical and intelligent.

[0024] Extract features from the collected eye images to obtain the blink frequency , pupil diameter change range , red bloodshot coverage ratio , eyelid ptosis angle , eye moistness score and other characteristic parameters; Substitute the above parameters into the eye fatigue scoring function Perform weighted summation, the formula is as follows:

[0025] in, Decibel is the weight coefficient obtained by model training, satisfying , used to balance the impact of various features on the score; The final score It is divided into multiple eye protection level intervals, forming evaluation levels such as "good", "mild fatigue", "moderate fatigue" and "severe fatigue", and corresponding eye protection recommendations are pushed according to the levels.

[0026] The image acquisition module further includes a light sensor for monitoring ambient light intensity. This sensor monitors changes in the user's ambient light intensity in real time and integrates data with the entire system. When the system detects that ambient light intensity exceeds a set threshold (e.g., too bright or too dim), the system automatically triggers a pre-set response mechanism. First, it prompts the user to adjust their eye conditions (e.g., turning on the lights or moving away from strong light sources) through voice prompts, pop-up notifications, or vibration alerts. Second, it automatically adjusts camera acquisition parameters, including but not limited to key parameters such as exposure time, gain, white balance, and contrast, to ensure that the captured images meet standard requirements for brightness, clarity, and color reproduction. This function not only effectively prevents image blur, overexposure, or underexposure caused by abnormal lighting conditions, but also significantly improves image quality stability, providing accurate and reliable raw data input for subsequent eye image recognition and AI analysis models, enhancing the scientific and accurate evaluation results. Furthermore, the module possesses a self-learning capability, optimizing parameters based on the lighting characteristics of the user's environment over time, further enhancing the intelligent and adaptable nature of image acquisition.

[0027] The image processing module possesses multi-level image optimization capabilities, utilizing a pre-trained model based on deep learning to intelligently process collected eye images. This process primarily involves image enhancement, alignment of key eye features, and deep fusion of multimodal features. In terms of image enhancement, the system intelligently repairs and optimizes image clarity, contrast, brightness, and noise, improving image quality and ensuring subsequent recognition accuracy. In terms of feature alignment, facial recognition and eye location algorithms automatically capture the eye region and geometrically correct for angular deviations and posture changes in the image, ensuring uniformity and standardization of the eye region. In the deep fusion stage, the system integrates skin color models, eyelid structure models, and age-specific eye changes to create a fusion model of features, enabling compatible processing across genders, ages, and demographics. Through this multi-level image optimization strategy, the system can adaptively handle a variety of complex data samples, avoiding recognition bias caused by individual differences. This significantly enhances the applicability, universality, and robustness of the assessment system across a wide range of populations, providing high-quality data support for subsequent eye health assessments.

[0028] The eye feature recognition module utilizes a multi-task convolutional neural network (MT-CNN) model, capable of simultaneously processing multiple eye feature tasks, enabling efficient image recognition and behavioral analysis. This module concurrently completes key tasks, including eye posture recognition, conjunctival bloodshot density detection, and blink behavior analysis, within a single model architecture. Eye posture recognition can be used to determine the user's gaze direction, eye opening width, and fatigue status; bloodshot density detection reflects the degree of eye congestion, serving as a warning signal for eye fatigue and dry eyes; and blink behavior analysis measures parameters such as blink frequency and duration to reflect the user's eye usage frequency and fatigue level. The MT-CNN model boasts powerful feature learning capabilities and real-time computing performance, enabling rapid and accurate eye feature extraction in complex backgrounds and diverse facial structures. Furthermore, the module supports automated image acquisition and analysis based on user-defined time periods (e.g., hourly or daily), continuously recording eye feature changes and forming a dynamic health profile. Through long-term tracking and trend modeling, the system can identify early signs of eye health problems in advance and provide users with scientific eye care advice and intervention measures, thus achieving truly personalized and intelligent eye health management.

[0029] The evaluation model analysis module incorporates a time series modeling mechanism to dynamically and continuously model trends in users' eye care behaviors and predict their health status. This module integrates image features collected from users at different time periods (such as blink frequency, changes in redness, and pupil constriction) with behavioral characteristics (such as daily eye use duration, frequency of electronic device use, and lighting environment records) to generate multi-dimensional time series data input. Using a recurrent neural network (RNN) as the core modeling tool, it performs deep learning and trend analysis on this time series data. RNNs are capable of capturing long-term dependencies in data, enabling them to identify potential abnormal behavioral trends or excessive eye use risks from a user's eye use patterns over time. For example, if the system detects an abnormally low blink frequency and consistently high eye use duration over multiple days, it can predict the risk of dry eye or visual fatigue, generate timely warnings, and deliver personalized eye care recommendations, such as mandatory rest reminders, lighting adjustment prompts, or eye behavior optimization solutions. This mechanism effectively improves the system's sensitivity and forward-looking judgment capabilities to changes in the user's eye protection status, realizes the transition from "passive response" to "active intervention", and significantly enhances the intelligence level of the evaluation system and user experience.

[0030] The feedback and suggestion output module, based on a knowledge graph and a rule engine, constructs an intelligent recommendation system for eye care scenarios. It can provide scientific and accurate intervention recommendations based on the user's individualized health assessment results. The system has a built-in rich eye care medical knowledge base, covering basic knowledge such as the physiological structure of the eye, common vision problems, fatigue and dry eye mechanisms, and integrates relevant national vision health standards and authoritative clinical guidance specifications to provide reliable medical support for the output of recommendations. By constructing an eye care-related knowledge graph, the system dynamically associates the user's historical image features, behavioral data, and evaluation results with knowledge nodes to achieve rapid matching of personalized recommendation paths. In terms of recommendation logic, the system integrates a rule engine to automatically trigger corresponding intervention strategies based on the user's current status (such as prolonged screen staring, abnormal ambient lighting, low blinking frequency, etc.), and outputs multimodal information forms including graphic descriptions, video demonstrations, and voice prompts to ensure the diversity and clarity of information communication. In addition, the module also supports linkage with mobile terminals or desktop devices (such as APP, web pages), automatically pushing eye protection suggestions, training tasks and behavioral reminders, improving user execution efficiency and daily compliance, and further enhancing the practicality of the system and human-computer interaction experience.

[0031] The system also includes a user identity management and data privacy protection module specifically designed to ensure comprehensive protection of user personal information, image data, and behavioral profile data throughout the assessment and feedback process. This module utilizes high-strength data encryption technologies, such as AES-256 symmetric encryption and RSA asymmetric encryption, to encrypt the storage and transmission of sensitive information, such as user-uploaded or real-time collected image data, blink rate, and eye activity records, preventing data leakage or unauthorized access over the network. Furthermore, the system supports anonymous user assessments, allowing users to use the eye care assessment service without binding their real-life identity, effectively alleviating initial privacy concerns. The module also features a fine-grained data authorization management mechanism, allowing users to customize authorization scopes, such as specifying which data can be used for model training and which data is restricted to local analysis, enhancing user control over their data. To further enhance security, the system also includes a built-in periodic data purge mechanism that automatically deletes historical data at a user-defined interval to prevent information risks from long-term accumulation. The design and operation of the entire module strictly adhere to national information security standards such as the Personal Information Protection Law and the Cybersecurity Law, as well as internationally accepted privacy protection protocols (such as GDPR), to ensure that while providing high-quality assessment services, user data privacy and system compliance and security are protected to the greatest extent possible.

[0032] The system integrates with a variety of third-party devices and platforms, enabling cross-device data collection and comprehensive analysis, thereby comprehensively improving the accuracy and practicality of eye care assessments. Specifically, the system seamlessly integrates with a variety of devices, including tablets, e-book readers, smartphones, and online learning platforms, to capture real-time data on users' eye behavior in different scenarios. This includes key information such as continuous eye use duration, screen brightness adjustment, and the type and frequency of reading content. Using this rich data input, the system can deeply analyze users' eye habits and environmental characteristics, identifying potential risk factors such as excessive eye use, inappropriate lighting, or poor reading posture. Furthermore, the cross-device collaborative monitoring mechanism enables the system to break down data silos on individual devices, enabling comprehensive tracking and dynamic assessment of users' eye behavior, improving the accuracy of the model and the scientific nature of personalized recommendations. More importantly, by integrating data from multiple platforms, the system can build a complete eye care management ecosystem, delivering eye care recommendations and reminders through multiple channels simultaneously, helping users develop healthy eye habits in different usage environments. This move not only enhances the practical value of the system, but also lays a solid foundation for creating intelligent and personalized vision health management solutions, and promotes eye protection technology towards multi-device collaboration and ecological development.

[0033] The system supports an advanced personalized learning mechanism, enabling continuous optimization based on the user's actual usage and health status. By continuously collecting user image analysis data and execution feedback, the system monitors subtle changes in the user's eye health and behavioral habits in real time, automatically adjusting the parameters of the internal model and the weighting of various evaluation indicators. This dynamic adjustment mechanism enables the evaluation system to not only adapt to individual differences among users but also evolve as their eye care behaviors and physiological conditions change. Through continuous model iteration and training, the system's prediction accuracy and adaptability are significantly improved, enabling it to more accurately reflect users' current eye care needs and potential health risks. Furthermore, this personalized learning mechanism ensures that the system effectively prevents model overfitting and performance degradation in the face of long-term data accumulation, maintaining stable and continuous evaluation capabilities. This intelligent, dynamically optimized evaluation method greatly meets the needs of personalized eye health management, providing users with scientific, personalized, and continuously effective health monitoring and guidance services, promoting the long-term maintenance and improvement of eye health.

[0034] An AI image analysis eye protection assessment method is used in conjunction with the above-mentioned AI image analysis eye protection assessment system; the method steps are as follows: Step 1: Image capture: High-definition cameras are used to capture real-time eye image information while the user is using a terminal device (such as a computer, tablet, or mobile phone). This ensures that the lighting, angle, and distance during image capture meet the set standards. Ambient light sensors are also used to simultaneously capture ambient brightness data, providing a reference for subsequent image analysis. Step 2: Image preprocessing: The raw image data is processed through operations such as denoising, color normalization, brightness compensation, and automatic eye area recognition and extraction to ensure the clarity and accuracy of image features. Image processing parameters are also personalized based on the user's background (such as skin color and glasses). Step 3: Feature extraction: Utilize a convolutional neural network model to extract deep features from eye images, including but not limited to blink frequency, pupil diameter, periocular bloodshot density, eyelid position and range of change. These features are then structured and encoded for subsequent model evaluation. Step 4: Assessment Modeling and Analysis: The extracted eye image features and historical behavior data are input into the eye protection assessment model. Based on a deep neural network and time series prediction mechanism, the model comprehensively analyzes the user's eye protection status, generates an eye protection risk level and trend chart, and provides a score for the impact of the user's current eye use behavior on vision health. Step 5. Result output and personalized recommendations: Generate personalized eye protection suggestions based on the evaluation results, and combine the user's pace of life, learning and work patterns, and environmental conditions to recommend improvement strategies such as eye time planning, rest frequency, and eye protection training methods. These suggestions are then pushed intelligently through the app, web page, or voice assistant to facilitate timely receipt and execution by users.

[0035] The system first uses high-definition cameras to capture real-time eye image information while the user is using a terminal device (such as a computer, tablet, or mobile phone). It also utilizes ambient light sensors to simultaneously monitor ambient brightness, ensuring that the lighting, angle, and distance during the acquisition process meet preset standards, laying a solid foundation for subsequent image analysis. The collected raw image data then undergoes preprocessing, including denoising, color normalization, brightness compensation, and automatic identification and extraction of the eye region. Image processing parameters are personalized based on the user's individual background information (such as skin color and eyeglasses worn) to ensure the clarity and accuracy of image features. Subsequently, a convolutional neural network model is used to extract deep features from the eye images, covering key indicators such as blink frequency, pupil diameter, periocular bloodshot density, eyelid position, and range of change. These features are then structured and encoded to facilitate subsequent model evaluation.

[0036] After feature extraction is completed, the system combines the eye image features with the user's historical behavior data, inputs them into the eye protection assessment model, and uses deep neural networks and time series prediction mechanisms to conduct a comprehensive analysis of the user's eye protection status, generate eye protection risk levels and trend charts, and assess the impact of current eye use behavior on vision health. Finally, based on the evaluation results, the system intelligently generates personalized eye protection recommendations. Combined with the user's pace of life, work and study patterns, and environmental conditions, it recommends scientific eye time planning, reasonable rest frequency, and effective eye protection training methods. All recommendations are intelligently pushed through multiple channels such as APP, web pages, or voice assistants to ensure that users can obtain them in a timely manner and easily implement them, thereby achieving all-round and dynamic eye health management.

[0037] Various modifications to the present disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the scope of the present disclosure. Therefore, the present disclosure is not limited to the examples and designs described herein, but should be given the widest scope consistent with the principles and novel features disclosed herein. Although one or more exemplary embodiments of the present disclosure have been described with reference to the accompanying drawings, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the present disclosure as defined in the appended claims.

Claims

1. An AI image analysis eye protection assessment system, characterized in that: It includes image acquisition module, image processing module, eye feature recognition module, evaluation model analysis module and feedback suggestion output module; The image acquisition module is used to collect user facial image information in real time through a camera device; The image processing module pre-processes the collected images, including image denoising, brightness adjustment, contrast enhancement, face location and eye area segmentation; The eye feature recognition module extracts features from processed images based on a deep learning algorithm; The evaluation model analysis module combines user historical data with current image analysis results to conduct a comprehensive evaluation of eye protection status by building a neural network model; The feedback and suggestion output module provides users with personalized eye protection suggestions based on the evaluation results; Extract features from the collected eye images to obtain the blink frequency , pupil diameter change range , red bloodshot coverage ratio , eyelid ptosis angle , eye moistness score and other characteristic parameters; Substitute the above parameters into the eye fatigue scoring function Perform weighted summation, the formula is as follows: in, Decibel is the weight coefficient obtained by model training, satisfying , used to balance the impact of various features on the score; The final score It is divided into multiple eye protection level intervals, forming evaluation levels such as "good", "mild fatigue", "moderate fatigue" and "severe fatigue", and corresponding eye protection recommendations are pushed according to the levels.

2. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The image acquisition module further includes a light sensor for monitoring the ambient light intensity and is linked to the system. When it detects that the ambient light is too strong or too dark, it automatically issues a prompt and adjusts the camera acquisition parameters to obtain data images that meet the standards, ensuring that image acquisition is carried out under reasonable lighting conditions, thereby improving the stability and accuracy of subsequent image recognition and analysis.

3. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The image processing module has multi-level image optimization capabilities, and realizes image enhancement, eye feature alignment and deep fusion based on pre-trained models. It can adaptively process data samples with different skin colors, age groups and eye structure differences, thereby meeting the eye protection assessment needs of different genders and age groups, and enhancing the universality and robustness of the system.

4. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The eye feature recognition module adopts a multi-task convolutional neural network model (MT-CNN) to simultaneously perform eye posture recognition, bloodshot density detection and blinking behavior analysis. It has real-time recognition and analysis capabilities, and can continuously record data according to the time period set by the user to achieve dynamic tracking of eye health status.

5. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The evaluation model analysis module introduces a time series modeling mechanism, integrates the image features and behavioral characteristics of users in different time periods, and models and predicts eye protection behavior trends based on a recurrent neural network (RNN). It can identify potential excessive eye use tendencies or bad behavior trends and provide intervention suggestions in advance.

6. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The feedback suggestion output module is based on the knowledge graph and rule engine, combined with the eye protection medical knowledge base and the national vision health standards, to intelligently recommend and match the user's individual conditions, output multimodal eye protection plans including text, images, voice, etc., and support automatic push to the user's mobile phone or computer APP, thereby improving user compliance and execution efficiency.

7. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The system also includes a user identity management and data privacy protection module, which is used to encrypt and store user images and behavioral data, support user anonymous evaluation, data authorization management and regular data clearing mechanism, protect user personal privacy and data security, and comply with relevant information security standards and privacy protection regulations.

8. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The system can be linked with third-party devices or platforms, such as tablets, e-book readers, online learning platforms, etc., to obtain real-time user eye usage time, screen brightness and reading content, and further improve assessment accuracy through cross-device collaborative monitoring and analysis to form a comprehensive eye protection management ecosystem.

9. The AI ​​image analysis eye protection assessment system according to claim 1, characterized in that: The system supports a personalized learning mechanism. By continuously collecting users' image analysis data and execution feedback, it automatically adjusts model parameters and evaluation weights to achieve an evaluation system that self-evolves with user changes. It also improves the accuracy and adaptability of long-term evaluations through model iteration to meet the needs of individuals for continuous eye health management.

10. An AI image analysis eye protection assessment method, the method being used in conjunction with an AI image analysis eye protection assessment system according to any one of claims 1 to 9; characterized in that: The steps of the method are as follows: Step 1: Image capture: High-definition cameras are used to capture real-time eye image information while the user is using a terminal device (such as a computer, tablet, or mobile phone). This ensures that the lighting, angle, and distance during image capture meet the set standards. Ambient light sensors are also used to simultaneously capture ambient brightness data, providing a reference for subsequent image analysis. Step 2: Image preprocessing: The raw image data is processed through operations such as denoising, color normalization, brightness compensation, and automatic eye area recognition and extraction to ensure the clarity and accuracy of image features. Image processing parameters are also personalized based on the user's background (such as skin color and glasses). Step 3: Feature extraction: Utilize a convolutional neural network model to extract deep features from eye images, including but not limited to blink frequency, pupil diameter, periocular bloodshot density, eyelid position and range of change. These features are then structured and encoded for subsequent model evaluation. Step 4: Assessment Modeling and Analysis: The extracted eye image features and historical behavior data are input into the eye protection assessment model. Based on a deep neural network and time series prediction mechanism, the model comprehensively analyzes the user's eye protection status, generates an eye protection risk level and trend chart, and provides a score for the impact of the user's current eye use behavior on vision health. Step 5. Result output and personalized recommendations: Generate personalized eye protection suggestions based on the evaluation results, and combine the user's pace of life, learning and work patterns, and environmental conditions to recommend improvement strategies such as eye time planning, rest frequency, and eye protection training methods. These suggestions are then pushed intelligently through the app, web page, or voice assistant to facilitate timely receipt and execution by users.

Images