Eye fatigue real-time detection method and device

Abstract

The invention provides an eyestrain real-time detection method and device, and relates to the technical field of eyestrain detection. The method comprises the following steps: identifying a current task state, and generating a task state feature; the method comprises the following steps: collecting eye data of a user, extracting eye behavior characteristics, and marking the eye behavior characteristics; by comparing the actual eye behavior characteristics with expected eye behaviors in the current task state, deviation is calculated, and relevance indexes are obtained; based on the relevance indexes and the eye features, the fatigue state is judged, and the fatigue level is obtained; and according to the fatigue level, changing the display element attribute of the current interaction so as to provide fatigue early warning. According to the method provided by the invention, eye fatigue detection can be realized, and fatigue early warning information is transmitted to the user in a low-interference and uninterrupted user concentration mode.

Description

Technical Field

[0001] The present invention relates to the technical field of eye fatigue detection, and in particular, to a method and device for real-time detection of eye fatigue. Background Art

[0002] A user wears a wearable device integrated with a real-time eye fatigue detection function, such as smart glasses, while performing tasks that require long-term continuous fine visual input and operation. The device internally contains sensors for collecting the user's eye data, such as a micro camera or an infrared sensor. The device processor receives the raw eye data stream collected by the sensors. In a specific working environment, such as when performing tasks like precision instrument assembly, long-term code writing, or remote medical image interpretation, the user needs to maintain a high level of concentration and observe and process details for a long time. In this specific task mode, the user needs to keep the visual focus fixed on a relatively small area or object for a long time and may need to make very fine eye movements to track or check details. This continuous visual load and the requirements for the eye accommodation system cause the eye muscles to be in a tense state for a long time. As the task continues and time passes, the user's eyes gradually accumulate fatigue. This eye fatigue will cause subtle changes physiologically and behaviorally. For example, the decline in the ciliary muscle accommodation ability may lead to worse focus stability, manifested as minute forward and backward focus drifts or accommodation lags; the weakening of the eye movement control ability may cause changes in the frequency or amplitude of minute eye tremors; long-term fixation may lead to a decrease in the blink frequency, but the eyelids may show slight drooping or incomplete closure. However, the fine operation tasks performed by the user themselves also require a high level of visual stability, precise eye movement control, and long-term continuous fixation. The eye behavior patterns caused by the nature of the task (such as long-term fixed fixation on a specific area, small-range precise saccades, suppression of non-task-related blinks) may be similar or confused with some physiological or behavioral changes that may be caused by early or moderate eye fatigue (such as prolonged fixation time, reduced saccades, decreased blink frequency). For example, the user does not blink for a long time in order to observe details, which is both a task requirement and may be a manifestation of the decline in the ability to inhibit the blink reflex due to fatigue. Minute eye tremors may be related to fatigue, but may also be physiological tremors generated when the user is trying to maintain precise fixation on a specific point, or related to minute involuntary head movements. Therefore, the eye fatigue detection method integrated in the wearable device needs to be able to distinguish the eye behavior patterns caused by the nature of the task and the user's active control from the involuntary physiological or behavioral changes caused by the accumulation of eye physiological fatigue. This requires the device to be able to not only collect eye surface feature data, but also analyze deeper eye physiological indicators, such as evaluating the state of the eye accommodation function through parameters such as the reaction speed of the pupil to different stimuli and the amount of accommodation lag, rather than relying solely on superficial blink or eye movement behaviors. In addition, when performing tasks that require a high level of concentration and fine operation, the user usually needs to maintain concentration, and any form of interference may affect the smooth progress of the task, and even cause serious consequences in some high-risk scenarios (such as precision surgery assistance, industrial hazardous material handling).Traditional eye fatigue warning methods, such as popping up text prompts in the center of the smart glasses display screen, emitting sound alarms, or the device generating obvious vibrations, may distract users and interrupt their operation processes. Therefore, wearable devices need to convey eye fatigue information to users in real time and accurately without disturbing the user's current delicate tasks. This may require exploring non-invasive and low-interference feedback mechanisms, such as displaying information in a very gentle and unobtrusive manner through the edge area of the smart glasses display screen, or using other modalities that do not rely on vision or hearing for prompting, but it is challenging to implement on resource-constrained wearable devices. At the same time, to ensure the accuracy of detection, the device needs to process high-resolution and high-frame-rate eye data streams, extract and analyze complex physiological and behavioral characteristics, and run a judgment model that can robustly distinguish between task effects and fatigue effects. These complex computational tasks need to achieve real-time response under the limited processing power and power consumption budget of the wearable device and ensure the timeliness and accuracy of the eye fatigue judgment results, which are the key technical problems that need to be overcome in integrating this function. Summary of the Invention

[0003] The purpose of the present invention is to provide a real-time eye fatigue detection method and device, which can effectively solve the technical problems faced by users when wearing wearable devices integrated with real-time eye fatigue detection functions during long-term continuous delicate visual input and operation tasks, realize eye fatigue detection, and convey fatigue warning information to users in a low-interference and non-interrupting manner for user concentration.

[0004] In the first aspect, the present invention provides a real-time eye fatigue detection method applied to a wearable device, including the following steps: By capturing the interaction events of the user in the task application, identifying the current task state, and generating task state features; By collecting the user's eye data, extracting eye behavior features, and annotating the eye behavior features according to the task state features; Establish a mapping between the task state and the expected eye behavior, calculate the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task state, and obtain a correlation index; Based on the correlation index and eye features, judge the fatigue state and obtain the fatigue level; According to the fatigue level, change the attribute of the display element of the current interaction to provide a fatigue warning.

[0005] The real-time eye fatigue detection method provided by the present invention uses task data to establish an "expected pattern" of eye behavior, compares the actual eye behavior with this expected pattern for analysis, identifies the deviation caused by fatigue, and conveys information to users in a low-interference manner.

[0006] Second aspect, the present invention provides a real-time eye fatigue detection device, which is applied to a wearable device and includes: A capture module, configured to identify the current task state by capturing the interaction events of the user in the task application and generate task state features; An acquisition module, configured to extract eye behavior features by collecting the user's eye data and label the eye behavior features according to the task state features; A comparison module, configured to establish a mapping between the task state and the expected eye behavior, calculate the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task state, and obtain a correlation index; A judgment module, configured to judge the fatigue state based on the correlation index and the eye features to obtain a fatigue level; A display module, configured to change the attribute of the display element of the current interaction according to the fatigue level to provide a fatigue warning.

[0007] As can be seen from the above, the real-time eye fatigue detection method provided by the present invention, when the user wears a wearable device integrated with the real-time eye fatigue detection function and performs a task that requires long-term continuous fine visual input and operation under limited computing resources and power consumption limitations, in a complex scenario where there is confusion in the behavioral performance between the high concentration and fine operation required by the task nature and the involuntary changes caused by eye physiological fatigue, it can distinguish in real time the eye behavior patterns caused by the task nature and the user's active control from the involuntary changes caused by the accumulation of eye physiological fatigue, and provide an accurate eye fatigue warning in a way that does not interfere with the user's current fine task.

[0008] Other features and advantages of the present invention will be described in the subsequent specification, and, in part, will be obvious from the specification, or will be understood by implementing the embodiments of the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the written specification and the drawings. Description of the Drawings

[0009] Figure 1 It is a flowchart of a real-time eye fatigue detection method provided by an embodiment of the present invention.

[0010] Figure 2 It is a schematic structural diagram of a real-time eye fatigue detection device provided by an embodiment of the present invention.

[0011] Label Description: 100, capture module; 200, acquisition module; 300, comparison module; 400, judgment module; 500, display module. Detailed Embodiments

[0012] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0013] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for descriptive distinction and cannot be understood as indicating or implying relative importance.

[0014] Referring to the Figure 1 drawings, the present invention provides a real-time eye fatigue detection method applied to a wearable device, including the following steps: By capturing the interaction events of the user in the task application, identifying the current task state, and generating task state features; By collecting the user's eye data, extracting eye behavior features, and annotating the eye behavior features according to the task state features; Establish a mapping between the task state and the expected eye behavior, and calculate the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task state to obtain a correlation index; Based on the correlation index and the eye features, judge the fatigue state to obtain the fatigue level; According to the fatigue level, change the attributes of the display elements of the current interaction to provide a fatigue warning.

[0015] Among them, capturing the task status of the user interaction event recognition task and generating task status features are achieved by monitoring the user's input operations or system events in the application, thereby determining the type and stage of the task the user is performing, and extracting parameters describing the task attributes. Collecting the user's eye data, extracting eye behavior features and annotating them according to the task status features, obtaining the original eye data stream through the sensors integrated in the device, processing the data to quantify behaviors such as blinking, eye movement, and pupil change, and associating and storing these behavior data with the task status features at the same moment. Establishing the mapping between the task status and the expected eye behaviors, comparing and calculating the deviation to obtain the correlation index, establishing the baseline of the eye behaviors in the non-fatigued state under different task statuses through a preset model or historical data, comparing the currently collected eye behaviors with this baseline, and quantifying the degree of difference. Judging the fatigue state based on the correlation index and eye features to obtain the fatigue level, taking the correlation index and other eye features reflecting physiological fatigue as inputs, and outputting the current fatigue degree of the user through an algorithm model. Changing the attributes of the display elements according to the fatigue level to provide a fatigue warning, adjusting the visual attributes of specific elements on the user interface according to the judged fatigue level, and thus giving a prompt to the user.

[0016] Specifically, this method solves the technical problem of real-time detection of eye fatigue on a wearable device considering the user's current task status. The method first identifies the task the user is performing by capturing the interaction events of the user in the task application and generates task status features. This step provides the context information required to understand the user's eye behaviors. Next, the method collects the user's eye data and extracts eye behavior features from it. The extracted eye behavior features are annotated according to the task status features generated in the first step. Then, the method establishes the mapping relationship between the task status and the eye behaviors that the user is expected to have in this task status. By comparing the actually collected and annotated eye behavior features with the expected eye behaviors in the current task status, calculating the deviation between the two, a correlation index is obtained. This index reflects how different the user's current eye behaviors are from the expected behaviors under normal circumstances of this task. Based on the calculated correlation index and other eye features, the method judges the user's fatigue state and determines the fatigue level. The correlation index provides information on abnormal behaviors in the task context. Combining with other eye features, it can be judged whether the behavior change is caused by fatigue. Finally, according to the judged fatigue level, the method changes the attributes of the display elements of the current interaction interface, thereby providing a fatigue warning to the user. This way uses the device display interface itself for prompting, and can transmit fatigue information without interrupting the user's task. By integrating the task status information and eye behavior analysis, this method can identify eye fatigue, especially in the case where the task has specific requirements for eye behaviors, thus overcoming the misjudgment problem that may occur when simply relying on eye behavior features for judgment, and realizing the real-time fatigue detection and warning function on the wearable device.

[0017] In some specific embodiments, the user wears smart glasses to perform precision assembly tasks. The smart glasses application captures the user's zooming and panning operations on the assembly drawing, identifies the task status as "precision visual assembly", and generates task status features, such as "high visual accuracy required", "small fixation area", "low expected blink frequency". At the same time, the built-in camera of the glasses captures the user's eye images and extracts eye behavior features, such as the actual blink frequency (e.g., 5 times per minute), the average fixation duration (e.g., 800 milliseconds), and the pupil diameter (e.g., 3.5 millimeters). These features are labeled with the "precision visual assembly" task status. The system internally stores the expected eye behavior baselines of non-fatigued users in the "precision visual assembly" task status, such as an expected blink frequency of 6 - 10 times per minute and an average fixation duration of 500 - 700 milliseconds. The system compares the actual eye behavior (5 blinks per minute, 800 - millisecond fixation) with the expected baseline, calculates the deviation, and obtains a correlation index (e.g., 0.7). This correlation index, together with other eye features such as the pupil diameter change rate, is input into the fatigue judgment model, and the model outputs the fatigue level (e.g., moderate fatigue). Based on the moderate fatigue level, the system determines that the adjustment type is color and generates adjustment parameters, such as adjusting the background color of non - critical interface elements from light gray (#F0F0F0) to soft beige (#FFF5E0). When the current interaction interface is rendered, the background color of non - critical elements becomes beige, providing a visual fatigue warning without disturbing the user's ongoing precision assembly operation.

[0018] In certain embodiments, the steps of collecting user eye data, extracting eye behavior features, and labeling the eye behavior features according to task status features include: According to the task status features, determine the types of eye behavior features related to the current task status and generate an eye behavior feature selection strategy; According to the eye behavior feature selection strategy, configure the eye data collection parameters of the wearable device, and the eye data collection parameters include the collection frequency and the image resolution; According to the configured eye data collection parameters, collect the user's eye data; According to the task status features and the collected user eye data, extract the corresponding eye behavior features and label the extracted eye behavior features with the task status features.

[0019] Specifically, when a user wears a wearable device to perform long-term fine visual tasks, the resources of the wearable device are limited. Collecting and processing all possible eye behavior characteristics leads to an increase in computational complexity and power consumption. At the same time, the correlation between eye behavior and specific task states is not clear, affecting the accuracy of fatigue judgment. To address this problem, this solution provides an optimized eye data collection and feature extraction process. First, the system determines the type of eye behavior characteristics most relevant to the current task state based on the currently identified task state characteristics, such as task type, the current operation focus of the user, etc. For example, for tasks that require long-term fixation and fine adjustment, the system may determine that characteristics such as fixation stability, pupil diameter change, and accommodation lag are more important; for tasks that require rapid saccades and information search, the system may determine that characteristics such as saccade amplitude, saccade speed, and blink frequency are more important. Thus, an eye behavior feature selection strategy is generated, which clarifies the types of eye data that need to be collected and analyzed with emphasis. Further, according to this feature selection strategy, the system dynamically configures the eye data collection parameters on the wearable device. For example, if rapid eye movement or microtremors need to be analyzed, the collection frequency is increased; if the pupil diameter or eyelid position needs to be accurately measured, the image resolution is increased. This configuration ensures that the collected data can effectively support the extraction of the subsequently selected features, avoiding the collection of irrelevant or low-value data, thereby reducing the computational complexity and power consumption of data processing. Then, the system collects the user's eye data stream according to the configured parameters. Finally, using the current task state characteristics and the collected eye data, the system executes a targeted feature extraction algorithm to extract only the eye behavior characteristics determined in the previous strategy. Each eye behavior feature data point or data segment extracted is labeled with the current task state characteristics. In this way, the eye behavior data is given clear task context information, enabling subsequent fatigue judgment to distinguish between eye behavior caused by task requirements and eye behavior caused by fatigue, improving the accuracy of judgment. Thus, this solution realizes the efficient and accurate acquisition of task-related eye behavior data on resource-constrained wearable devices.

[0020] In some specific embodiments, the user wears smart glasses to perform precision electronic component soldering tasks. The capture module identifies the current task status as "precision soldering" and generates task status features, such as "task type: soldering" and "operation object: tiny components". Based on these task status features, the system determines the types of eye behavior features related to precision soldering, including fixation point stability, pupil diameter, blink frequency, and eyelid closure degree. Thereby, an eye behavior feature selection strategy is generated. According to this strategy, the acquisition module configures the data acquisition parameters of the eye camera. For example, the acquisition frequency is set to 120 Hz to capture tiny eye movements, and the image resolution is set to 720p to accurately measure the pupils and eyelids. The acquisition module acquires the user's eye video stream according to the configured parameters. Subsequently, based on the "precision soldering" task status features and the acquired video data, features such as the fixation point coordinate sequence, pupil diameter value, blink event count, and eyelid height are extracted. Each set of extracted feature data, such as the average fixation point drift amplitude or average pupil diameter calculated within a certain time period, is labeled with "task status: precision soldering". Thus, the subsequent comparison module and judgment module can utilize these eye behavior features with task context information for fatigue analysis.

[0021] In certain embodiments, the steps of establishing the mapping between the task status and the expected eye behavior, calculating the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task status, and obtaining the correlation index include: Obtain the user's historical task data, where the historical task data includes the task type, task completion time, and task operation frequency; According to the task type, perform a weighted average on the historical task data to calculate the proficiency score of the user for the current task; According to the proficiency score, generate an adjustment parameter using a Gaussian function; the higher the proficiency score, the larger the adjustment parameter; According to the adjustment parameter, expand the mapping range between the task status and the expected eye behavior to generate a personalized mapping relationship; According to the personalized mapping relationship, compare the actual eye behavior features with the expected eye behavior in the current task status, calculate the Euclidean distance between the actual eye behavior and the expected eye behavior, and obtain the correlation index.

[0022] Specifically, this solution aims to address the problem that the individual differences of users, especially the differences in task proficiency, are not considered when establishing the mapping between task states and expected eye behaviors, resulting in inaccurate correlation metrics. This solution evaluates the user's proficiency in the current task by introducing the user's historical task data and makes personalized adjustments to the mapping between task states and expected eye behaviors based on this. First, historical data of the user's past execution of similar tasks is obtained, which reflects the user's task completion efficiency and operation habits. Then, the historical data is weighted according to the task type to calculate a proficiency score, which quantifies the user's experience and ability in a specific task. Next, the Gaussian function is used to convert the proficiency score into an adjustment parameter. The higher the proficiency, the larger the generated adjustment parameter, indicating that proficient users may exhibit a greater range of individual differences or more efficient patterns in eye behaviors. Using this adjustment parameter, the mapping range between task states and expected eye behaviors is dynamically expanded or adjusted, thereby generating a personalized mapping relationship that better matches the individual proficiency level of the user. This personalized mapping relationship is no longer a fixed standard but an expected behavior range adjusted according to the user's actual ability. Finally, based on the personalized mapping relationship, the Euclidean distance between the actually collected eye behavior features and the user's personalized expected eye behaviors in the current task state is calculated. This Euclidean distance serves as a correlation metric, more accurately reflecting whether the user's current eye behavior deviates from its normal or efficient pattern based on proficiency, thus providing a more reliable basis for subsequent fatigue judgment. In this way, this solution can distinguish the normal behavior differences caused by task proficiency from the behavior changes caused by fatigue, improving the accuracy of eye fatigue detection.

[0023] In some specific embodiments, for example, assume that the user is performing a "code writing" task. The system obtains the historical data of the user's past "code writing" tasks, including the completion time and operation frequency of each task (e.g., the number of keystrokes, the number of mouse clicks). Based on this historical data, the proficiency score of the user in the "code writing" task is calculated as 0.8 (score range 0 - 1) through weighted average. Using the Gaussian function f(x)=a*exp(-(x - b)^2 / (2c^2)), where x is the proficiency score and a, b, c are preset parameters, the adjustment parameter is calculated as 1.5. Assume that for the "code writing" task, the general expected blink frequency range is 15 - 20 times per minute. Using the adjustment parameter 1.5, this range is expanded. For example, the personalized expected blink frequency range becomes 15 - 30 times per minute, where 20 * 1.5 = 30. The system real-time collects the user's current actual blink frequency as 22 times per minute. The actual blink frequency (22) is compared with the personalized expected range (15 - 30), and the Euclidean distance is calculated. For example, the distance between the actual value and the range boundary, or the distance from the range center, can be calculated. The obtained Euclidean distance is the correlation index, which reflects the degree to which the user's current blink behavior deviates from its personalized expected range and is used for subsequent fatigue judgment.

[0024] In certain embodiments, the steps of changing the attributes of the display elements of the current interaction according to the fatigue level to provide a fatigue warning include: Determine the adjustment type of the display element attributes according to the fatigue level; Generate the adjustment parameters of the display element attributes according to the fatigue level and the adjustment type. The higher the fatigue level, the greater the amplitude of the adjustment parameters; Adjust the background color, font color, and border color of the display elements of the current interaction interface according to the adjustment parameters; Provide a fatigue warning by rendering the current interaction interface according to the adjusted display element attributes.

[0025] Furthermore, the adjustment types include color, transparency, and blink frequency.

[0026] Among them, according to the fatigue level, the adjustment type of the display element attributes is determined. This step selects a visual adjustment strategy based on the received fatigue level. For example, when the fatigue level is low, color adjustment can be selected; when the fatigue level is high, blink frequency adjustment can be selected. Further, according to the fatigue level and the adjustment type, the adjustment parameters of the display element attributes are generated. This step calculates the adjustment amplitude, and the higher the fatigue level, the greater the amplitude of the adjustment parameters. For example, for color adjustment, the parameter can be a color offset; for blink frequency adjustment, the parameter can be a Hertz value. Thus, according to the adjustment parameters, the background color, font color, and border color of the display elements in the current interaction interface are adjusted. This step applies the calculated parameters to the corresponding attributes of the user interface elements. Specifically, these attribute values can be modified by accessing the device's graphics rendering interface or UI framework. Finally, according to the adjusted display element attributes, by rendering the current interaction interface, a fatigue warning is provided. This step presents the modified interface to the user, conveying the fatigue information through visual changes. This process realizes the integration of fatigue information into the user interface, reducing the interference with the user's tasks.

[0027] Specifically, this technical solution receives the fatigue level output by the eye fatigue detection method. This fatigue level is used to determine how to adjust the display element attributes of the user interface. First, the system selects an adjustment type according to the fatigue level, such as color change or transparency change. Then, according to the selected adjustment type and the current fatigue level, specific adjustment parameters are calculated. The higher the fatigue level, the larger the calculated adjustment parameter value, making the visual change more obvious. Then, the system uses these adjustment parameters to modify the attributes of specific display elements in the current user interface, such as background color, font color, or border color. For example, as the fatigue level increases, the background color may gradually change to a warm color tone, or the border color becomes more prominent. Finally, the device re-renders the user interface, presenting the display elements with modified attributes to the user. The user perceives the fatigue state by observing the visual changes of the interface elements. This way integrates the fatigue warning information into the task interface that the user is performing, avoiding the interruption that may be caused by pop-up windows or sound prompts, and at the same time realizing a differentiated warning intensity through the amplitude of the adjustment parameters, making the warning information match the fatigue degree.

[0028] In some specific embodiments, when it is detected that the user is in a state of moderate fatigue, the system determines that the adjustment type is color adjustment. Based on the moderate fatigue level, the system generates a color adjustment parameter, for example, adjusting the border color of the main operation button in the current interactive interface from gray to light orange, and the adjustment range corresponds to the moderate fatigue level. Further, the system applies the light orange value to the border properties of these buttons. Thus, when the interface is rendered, the user sees that the border of the operation button is light orange, thereby receiving a visual warning of moderate fatigue. As another preferred embodiment, when it is detected that the user is in a state of high fatigue, the system determines that the adjustment type is a flickering frequency adjustment. Based on the high fatigue level, the system generates a flickering frequency parameter, for example, flickering once per second. This parameter is applied to the background elements of a non-core area in the interface. Thus, the background element flickers at a frequency of once per second, providing a stronger visual warning to prompt the user of high fatigue. These adjustments are all integrated into the current task interface and do not generate additional interruption elements.

[0029] Reference Attachment Figure 2 The present invention provides a real-time eye fatigue detection device, which is applied to wearable devices and includes: The capture module 100 is used to identify the current task state and generate task state features by capturing user interaction events in the task application; The acquisition module 200 is used to collect user eye data, extract eye behavior features, and label the eye behavior features according to task state features; Comparison module 300 is used to establish a mapping between task status and expected eye behavior, and calculate the deviation by comparing the actual eye behavior characteristics with the expected eye behavior under the current task status to obtain a correlation index; The judgment module 400 is used to judge the fatigue state based on the correlation index and eye characteristics and obtain the fatigue level; The display module 500 is used to change the attributes of the currently interacted display elements according to the fatigue level to provide fatigue warning.

[0030] In some embodiments, the collection module 200 collects user eye data, extracts eye behavior features, and labels the eye behavior features according to task state features, and performs the following operations: According to the task state characteristics, determine the eye behavior feature type related to the current task state and generate an eye behavior feature selection strategy; According to the eye behavior feature selection strategy, configure the wearable device's eye data collection parameters, including collection frequency and image resolution; Collect user eye data according to the configured eye data collection parameters; According to the task status features and the collected user eye data, extract the corresponding eye behavior features, and label the extracted eye behavior features using the task status features.

[0031] In some embodiments, the comparison module 300 performs the following when establishing the mapping between the task status and the expected eye behavior, calculating the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task status, and obtaining the correlation index: Obtain the user's historical task data, where the historical task data includes the task type, task completion time, and task operation frequency; According to the task type, perform a weighted average on the historical task data to calculate the proficiency score of the user for the current task; According to the proficiency score, generate an adjustment parameter using a Gaussian function; the higher the proficiency score, the larger the adjustment parameter; According to the adjustment parameter, expand the mapping range between the task status and the expected eye behavior to generate a personalized mapping relationship; According to the personalized mapping relationship, compare the actual eye behavior features with the expected eye behavior in the current task status, calculate the Euclidean distance between the actual eye behavior and the expected eye behavior, and obtain the correlation index.

[0032] In some embodiments, the display module 500 performs the following when changing the display element attributes of the current interaction according to the fatigue level to provide a fatigue warning: Determine the adjustment type of the display element attributes according to the fatigue level; According to the fatigue level and the adjustment type, generate an adjustment parameter for the display element attributes, and the higher the fatigue level, the greater the amplitude of the adjustment parameter; According to the adjustment parameter, adjust the background color, font color, and border color of the display elements on the current interaction interface; According to the adjusted display element attributes, provide a fatigue warning by rendering the current interaction interface.

[0033] In some embodiments, the adjustment types include color, transparency, and blink frequency.

[0034] In this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0035] The above are only embodiments of the present invention and are not used to limit the protection scope of the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A real-time eye fatigue detection method applied to a wearable device, characterized in that, It includes the following steps: By capturing the interaction events of the user in the task application, identifying the current task state, and generating task state features; By collecting the user's eye data, extracting eye behavior features, and annotating the eye behavior features according to the task state features; Establish a mapping between the task state and the expected eye behavior, calculate the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task state, and obtain a correlation index; Based on the correlation index and eye features, judge the fatigue state and obtain the fatigue level; According to the fatigue level, change the attribute of the display element of the current interaction to provide a fatigue warning.

2. The real-time eye fatigue detection method according to claim 1, wherein The step of collecting the user's eye data, extracting eye behavior features, and annotating the eye behavior features according to the task state features includes: According to the task state features, determine the type of eye behavior features related to the current task state and generate an eye behavior feature selection strategy; According to the eye behavior feature selection strategy, configure the eye data collection parameters of the wearable device, and the eye data collection parameters include the collection frequency and the image resolution; According to the configured eye data collection parameters, collect the user's eye data; According to the task state features and the collected user's eye data, extract the corresponding eye behavior features, and use the task state features to annotate the extracted eye behavior features.

3. The real-time eye fatigue detection method according to claim 1, characterized in that The step of establishing a mapping between the task state and the expected eye behavior, calculating the deviation by comparing the actual eye behavior features with the expected eye behavior in the current task state, and obtaining a correlation index includes: Obtain the user's historical task data, and the historical task data includes the task type, the task completion time, and the task operation frequency; According to the task type, perform a weighted average on the historical task data to calculate the proficiency score of the user for the current task; According to the proficiency score, generate an adjustment parameter using a Gaussian function; the higher the proficiency score, the larger the adjustment parameter; According to the adjustment parameter, expand the mapping range between the task state and the expected eye behavior to generate a personalized mapping relationship; According to the personalized mapping relationship, compare the actual eye behavior features with the expected eye behavior in the current task state, calculate the Euclidean distance between the actual eye behavior and the expected eye behavior, and obtain a correlation index.

4. The real-time eye fatigue detection method according to claim 1, wherein, The step of changing the attribute of the display element of the current interaction according to the fatigue level to provide a fatigue warning includes: According to the fatigue level, determine the adjustment type of the display element attribute; According to the fatigue level and the adjustment type, generate an adjustment parameter for the display element attribute, and the higher the fatigue level, the greater the amplitude of the adjustment parameter; According to the adjustment parameter, adjust the background color, font color, and border color of the display element of the current interaction interface; According to the adjusted display element attribute, provide a fatigue warning by rendering the current interaction interface.

5. The real-time eye fatigue detection method according to claim 4, wherein The adjustment types include color, transparency, and blink frequency.

6. A real-time eye fatigue detection device, applied to a wearable device, characterized in that, It includes: A capture module for capturing the interaction events of the user in the task application, identifying the current task state, and generating task state features; A collection module for collecting the user's eye data, extracting eye behavior features, and annotating the eye behavior features according to the task state features; A comparison module, configured to establish a mapping between task states and expected eye behaviors, calculate a deviation by comparing actual eye behavior features with the expected eye behaviors under the current task state, and obtain a correlation index; A judgment module, configured to judge the fatigue state based on the correlation index and eye features, and obtain a fatigue level; A display module, configured to change the attribute of a display element of the current interaction according to the fatigue level, so as to provide a fatigue warning.

7. The real-time eye fatigue detection device according to claim 6, characterized in that The acquisition module is executed when collecting user eye data, extracting eye behavior features, and labeling the eye behavior features according to task state features: Determine the type of eye behavior features related to the current task state according to the task state features, and generate an eye behavior feature selection strategy; Configure the eye data acquisition parameters of the wearable device according to the eye behavior feature selection strategy, where the eye data acquisition parameters include acquisition frequency and image resolution; Collect user eye data according to the configured eye data acquisition parameters; Extract corresponding eye behavior features according to the task state features and the collected user eye data, and label the extracted eye behavior features by using the task state features.

8. The real-time eye fatigue detection device according to claim 6, wherein The comparison module is executed when establishing a mapping between task states and expected eye behaviors, calculating a deviation by comparing actual eye behavior features with the expected eye behaviors under the current task state, and obtaining a correlation index: Obtain the user's historical task data, where the historical task data includes task type, task completion time, and task operation frequency; Perform weighted averaging on the historical task data according to the task type, and calculate the proficiency score of the user for the current task; Generate an adjustment parameter by using a Gaussian function according to the proficiency score; The higher the proficiency score, the larger the adjustment parameter; Expand the mapping range between the task state and the expected eye behavior according to the adjustment parameter, and generate a personalized mapping relationship; Compare the actual eye behavior features with the expected eye behaviors under the current task state according to the personalized mapping relationship, calculate the Euclidean distance between the actual eye behavior and the expected eye behavior, and obtain a correlation index.

9. The real-time eye fatigue detection device according to claim 6, characterized in that, The display module is executed when changing the attribute of a display element of the current interaction according to the fatigue level, so as to provide a fatigue warning: Determine the adjustment type of the display element attribute according to the fatigue level; Generate an adjustment parameter for the display element attribute according to the fatigue level and the adjustment type, and the higher the fatigue level, the greater the amplitude of the adjustment parameter; Adjust the background color, font color, and border color of the display elements of the current interaction interface according to the adjustment parameter; Provide a fatigue warning by rendering the current interaction interface according to the adjusted display element attributes.

10. The real-time eye fatigue detection device according to claim 9, characterized in that, The adjustment types include color, transparency, and blink frequency.

Images