Amblyopia training system based on target search

Abstract

Embodiments of the present disclosure disclose a target search-based amblyopia training system. The system comprises a brightness level configuration server, a personalized image generation server, an amblyopia training server and a user terminal device, wherein the brightness level configuration server is configured to determine a training brightness level; the personalized image generation server is configured to generate a test image set; the amblyopia training server is configured to control the user terminal device to perform amblyopia eye training operations according to the training contralateral healthy eye area and amblyopia eye area, and obtain amblyopia eye training results; change the training difficulty level according to the amblyopia eye training results; take the updated training difficulty level as the training difficulty level of the next test image, and continue to perform the amblyopia eye training step to obtain various amblyopia eye training results. The personalized implementation reduces the computing resources and storage resources consumed by amblyopia training, and improves the efficiency and training effect of amblyopia training.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the interdisciplinary field of image processing and visual training, specifically to a weak vision training system based on target search. Background Technology

[0002] Amblyopia training systems can generate test images for improving visual function in amblyopic individuals, enabling users to undergo visual function recovery training. Currently, existing amblyopia training systems typically employ the following method: occluding the user's contralateral healthy eye to force the user to observe using the amblyopic eye, and combining this with fine visual acuity training activities such as beading and drawing to promote the improvement of visual function in the amblyopic eye.

[0003] However, in practice, it has been found that when using the above methods to train users for amblyopia, the following technical problems often arise: Existing systems typically use fixed training parameters (such as fixed occlusion time and fixed training content), failing to consider individual differences among users in terms of age, gender, and interests. This results in inconsistent training effects, with some users requiring longer training periods to achieve visual function improvement. The system also needs to generate and store more training images and record more training data, which not only wastes a lot of computing and storage resources but also significantly increases server load, leading to longer overall training time and poorer training results.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide amblyopia training systems, methods, electronic devices, and computer-readable media based on target search to address one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide an amblyopia training system based on target search, comprising: a brightness level configuration server, a personalized image generation server, an amblyopia training server, and a user terminal device that are interconnected. The brightness level configuration server is used to determine a training brightness level based on a preset brightness level set and an acquired calibration image set. The personalized image generation server is used to generate a test image set based on user account information and a background material set. The amblyopia training server is used to perform the following amblyopia training steps based on each test image in the test image set and a training difficulty level: performing irregular segmentation processing on the test images to obtain a target level for amblyopia training. The system identifies two regions: the contralateral healthy eye region and the amblyopic eye region. Based on the training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. Based on the training contralateral healthy eye region and the amblyopic eye region, the user terminal device is controlled to perform amblyopic eye training operations to obtain amblyopic eye training results. Based on the amblyopic eye training results, the training difficulty level is modified to obtain an updated training difficulty level. The updated training difficulty level is used as the training difficulty level for the next test image, and the amblyopic eye training steps continue. In response to the detection that the number of executions of the amblyopic eye training steps meets a preset number condition, the amblyopic eye training is terminated, and each amblyopic eye training result is obtained.

[0008] Secondly, some embodiments of this disclosure provide a target-search-based amblyopia training method, including: determining a training brightness level based on a preset brightness level set and an acquired calibration image set; generating a test image set based on user account information and a background material set; and performing the following amblyopia training steps based on each test image in the test image set and a training difficulty level: performing irregular segmentation processing on the test images to obtain a contralateral healthy eye region and an amblyopia region; modifying the contralateral healthy eye region according to the training brightness level to obtain a training contralateral healthy eye region; controlling a user terminal device to perform amblyopia training operations based on the training contralateral healthy eye region and the amblyopia region to obtain an amblyopia training result; modifying the training difficulty level according to the amblyopia training result to obtain an updated training difficulty level; using the updated training difficulty level as the training difficulty level for the next test image and continuing to perform the above amblyopia training steps; and terminating the amblyopia training in response to detecting that the number of executions of the above amblyopia training steps meets a preset number condition, thereby obtaining various amblyopia training results.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in any of the implementations of the first or second aspect.

[0011] The above-described embodiments of this disclosure have the following beneficial effects: A target-search-based amblyopia training system according to some embodiments of this disclosure can reduce the computational and storage resources consumed in amblyopia training, lower server load, shorten the overall system training time, and improve the effectiveness of amblyopia training. Specifically, the reasons for the high computational and storage resources consumed in amblyopia training, high server load, long overall system training time, and poor training effect are as follows: Existing systems typically use fixed training parameters (such as fixed occlusion time and fixed training content), failing to consider individual differences among users in terms of age, gender, and interests, resulting in inconsistent training effects. Some users require longer training periods to achieve visual function improvement. The system needs to generate and store more training images and record more training data, which not only wastes a large amount of computational and storage resources but also significantly increases server load, leading to a long overall system training time and poor training effect. Based on this, some embodiments of the target search-based amblyopia training system disclosed herein include: a brightness level configuration server, a personalized image generation server, an amblyopia training server, and a user terminal device that are interconnected. The brightness level configuration server is used to determine the training brightness level based on a preset brightness level set and an acquired calibration image set. Thus, the brightness level for amblyopia training can be obtained. The personalized image generation server is used to generate a test image set based on user account information and a background material set. Thus, a test image set for amblyopia training can be obtained. The amblyopia training server is used to perform the following amblyopia training steps based on each test image in the test image set and the training difficulty level: First, the test images are irregularly segmented to obtain the contralateral healthy eye region and the amblyopia region. Thus, the segmented contralateral healthy eye region and the amblyopia region can be obtained. Then, based on the training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. Thus, the training contralateral healthy eye region for amblyopia training can be obtained. Next, based on the training regions of the contralateral healthy eye and the amblyopic eye, the user terminal device is controlled to perform amblyopic eye training operations to obtain amblyopic eye training results. This allows control over the user terminal device to perform amblyopic eye training operations. Then, based on the amblyopic eye training results, the training difficulty level is modified to obtain an updated training difficulty level. This updated training difficulty level is then used as the training difficulty level for the next test image, and the amblyopic eye training steps continue. Finally, in response to the detection that the number of executions of the amblyopic eye training steps meets a preset condition, the amblyopic eye training is terminated, and each amblyopic eye training result is obtained. This completes one round of amblyopic eye training.Because it doesn't use fixed training parameters, but instead determines personalized training brightness levels through a brightness level configuration server, the training brightness levels can adapt to the binocular inhibition levels of different users, reducing the number of invalid training sessions caused by brightness level mismatches, thereby reducing the server's computational load. Also, because it generates personalized test image sets through a personalized image generation server and performs amblyopia training based on these personalized test image sets, it reduces the waste of generated training images and allocated computing resources caused by users quitting midway due to boredom. Furthermore, because the amblyopia training server dynamically adjusts the training difficulty level based on the gaze type in the amblyopia training results, it adjusts the training difficulty level based on the user's actual gaze behavior quality, reducing repetitive and invalid training caused by inaccurate difficulty adjustments. This reduces the number of training images and training data that the system needs to generate and store, thus shortening the overall system training time and improving the amblyopia training effect. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a schematic diagram of the structure of some embodiments of the target search-based amblyopia training system according to the present disclosure; Figure 2 This is a timing diagram of some embodiments of an amblyopia training system based on object search according to the present disclosure; Figure 3 This is a flowchart of some embodiments of the weak vision training method based on object search according to the present disclosure; Figure 4 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure; Figure 5 This is an internal test diagram of a user-end device suitable for implementing this disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Figure 1 A schematic diagram of the structure of some embodiments of the target search-based amblyopia training system according to the present disclosure is shown.

[0021] like Figure 1As shown, the target search-based amblyopia training system provided in this disclosure may include: a brightness level configuration server 101, a personalized image generation server 102, an amblyopia training server 103, and a user terminal device 104. The brightness level configuration server 101 is a server used to determine the training brightness level based on a preset brightness level set and an acquired calibration image set. For example, the brightness level configuration server 101 may be a Dell PowerEdge R750. The personalized image generation server 102 is a server used to generate a test image set based on user account information and a background material set. For example, the personalized image generation server 102 may be an NVIDIA DGX A100. The aforementioned amblyopia training server 103 may be configured to perform the following amblyopia training steps based on each test image and training difficulty level in the aforementioned test image set: irregularly segmenting the test image to obtain a contralateral healthy eye region and an amblyopia region; modifying the contralateral healthy eye region according to the aforementioned training brightness level to obtain a training contralateral healthy eye region; controlling the user terminal device to perform amblyopia training operations based on the aforementioned training contralateral healthy eye region and the aforementioned amblyopia region to obtain an amblyopia training result; modifying the aforementioned training difficulty level according to the aforementioned amblyopia training result to obtain an updated training difficulty level; using the updated training difficulty level as the training difficulty level for the next test image and continuing to execute the aforementioned amblyopia training steps; and terminating the aforementioned amblyopia training in response to detecting that the number of executions of the aforementioned amblyopia training steps meets a preset number condition, thereby obtaining each amblyopia training result. For example, the amblyopia training server 103 may be an HPE ProLiant DL380 Gen11. The aforementioned user terminal device 104 may be a BOE (Beijing Oriental Electronics) far-image light screen used for amblyopia training. It should be noted that the aforementioned communication connections may include, but are not limited to, 3G / 4G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wideband) connections, and other currently known or future-developed communication methods.

[0022] The following is for reference. Figure 2 The diagram shows a timing diagram of a weak vision training system based on object search according to this disclosure.

[0023] like Figure 2 As shown, a target-search-based amblyopia training system includes: a brightness level configuration server, a personalized image generation server, an amblyopia training server, and a user terminal device. The interaction steps between the brightness level configuration server, the personalized image generation server, the amblyopia training server, and the user terminal device may include the following steps: In some embodiments, the brightness level configuration server can be configured to perform the following steps: Step 201: The brightness level configuration server is used to determine the training brightness level based on the preset brightness level set and the acquired calibration image set.

[0024] In some embodiments, the brightness level configuration server can determine the training brightness level based on a preset brightness level set and an acquired calibration image set. The preset brightness levels in the preset brightness level set can represent the brightness levels of the user terminal device. The user terminal device can represent a device used for binocular vision. Here, the specific type of the user terminal device is not limited; for example, the user terminal device can be a distance learning screen. The brightness level can represent a first brightness level, a second brightness level, a third brightness level, or a fourth brightness level. The first brightness level can represent 100% original brightness. The second brightness level can represent 75% original brightness. The third brightness level can represent 50% original brightness. The fourth brightness level can represent 25% original brightness. The original brightness can represent the original display brightness of the user terminal device. The calibration images in the calibration image set can represent images containing a target object used to determine the training brightness level. Here, the specific type of the target object is not limited; for example, the target object can be a white car. The above-mentioned training brightness levels can represent the brightness levels used for performing amblyopia training operations or healthy eye training operations.

[0025] In some optional implementations of certain embodiments, the brightness level configuration server described above can determine the training brightness level based on a preset brightness level set and the acquired calibration image set through the following steps: In addressing the technical problems mentioned above, and considering the application scenario of amblyopia training for children with severe anisometropic amblyopia, the following technical problem often arises: Significant differences in binocular inhibition levels among users mean that fixed training brightness levels cannot meet individual needs. This leads to frequent timeouts for some users due to excessively high brightness levels, or ineffective training due to excessively low brightness levels. The system needs to repeatedly execute invalid training sessions, recording a large amount of failure data. This not only wastes significant computing and storage resources but also greatly increases server load, resulting in longer overall training times and poorer training effects. To address the following requirements for this application scenario: adapting to differences in binocular inhibition levels among different users and adapting to changes in binocular inhibition levels for the same user at different training stages, we have decided to adopt the following personalized solution: The first step is to perform the following steps for each preset brightness level in the above set of preset brightness levels: First, for each calibration image in the above calibration image set, perform the following steps: The first sub-step involves generating a calibrated contralateral healthy eye region and a calibrated amblyopic eye region corresponding to the calibration image. The calibrated contralateral healthy eye region represents the visual field of view allocated to the contralateral healthy eye of the target object after irregular segmentation of the calibration image. The target object represents the user using the aforementioned user terminal device. The calibrated amblyopic eye region represents the visual field of view, including the target object, allocated to the amblyopic eye of the target object after irregular segmentation of the calibration image. In practice, the brightness level configuration server first uses a threshold segmentation algorithm to irregularly segment the calibration image to obtain the corresponding contralateral healthy eye region and amblyopic eye region as the calibrated contralateral healthy eye region and calibrated amblyopic eye region. It should be noted that the area of ​​the calibrated contralateral healthy eye region and the area of ​​the calibrated amblyopic eye region each occupy 50% of the area of ​​the calibration image.

[0026] The second sub-step involves determining the target brightness attenuation rate for each pixel in the calibrated contralateral healthy eye region based on the aforementioned preset brightness levels. The target brightness attenuation rate characterizes the proportion by which the brightness of each pixel in the calibrated contralateral healthy eye region needs to be reduced. In practice, the brightness level configuration server can convert the preset brightness levels into corresponding target brightness attenuation rates using a proportional mapping algorithm. For example, a target brightness attenuation rate of 1.0 corresponds to a preset brightness level of 100%, 0.75 to 75%, 0.5 to 50%, and 0.25 to 25%.

[0027] The third sub-step involves updating the calibrated contralateral healthy eye region based on the target brightness attenuation rate, resulting in a modified calibrated contralateral healthy eye region. This modified calibrated contralateral healthy eye region represents the field of view obtained after adjusting the brightness level of the calibrated contralateral healthy eye region to the preset brightness level. In practice, firstly, the brightness level configuration server can perform point-by-point multiplication to multiply the original brightness value of each pixel in the calibrated contralateral healthy eye region by the target brightness attenuation rate, obtaining various reduced brightness values. Then, through memory write operations, these reduced brightness values ​​are written to the display buffer at the positions corresponding to the pixels in the calibrated contralateral healthy eye region, thereby updating the calibrated contralateral healthy eye region and obtaining the modified calibrated contralateral healthy eye region. The original brightness value represents the initial brightness value of each pixel in the calibrated contralateral healthy eye region before brightness adjustment. The reduced brightness value represents the brightness value obtained by multiplying the original brightness value of each pixel in the calibrated contralateral healthy eye region by the target brightness attenuation rate.

[0028] The fourth sub-step involves determining a set of boundary pixels based on the aforementioned modified and calibrated contralateral healthy eye region and the aforementioned calibrated amblyopic eye region. The boundary pixels in this set represent pixels on the boundary line between the modified and calibrated contralateral healthy eye region and the aforementioned calibrated amblyopic eye region. In practice, firstly, the brightness level configuration server uses an eight-neighbor edge tracking algorithm to traverse all pixels in both regions and identifies pixels that meet preset criteria as boundary pixels. These preset criteria can be that the pixel belongs to either the modified and calibrated contralateral healthy eye region or the calibrated amblyopic eye region, and that at least one adjacent pixel belonging to the other region exists within the pixel's eight-neighbor domain. Then, the resulting set of boundary pixels is defined as the boundary pixel set.

[0029] The fifth sub-step involves feathering the boundaries of the modified calibration contralateral healthy eye region and the calibrated amblyopic eye region based on the aforementioned set of boundary pixels, resulting in a feathered calibration contralateral healthy eye region and a feathered calibration amblyopic eye region. The feathered calibration contralateral healthy eye region represents the visual field obtained after feathering the modified calibration contralateral healthy eye region. The feathered calibration amblyopic eye region represents the visual field obtained after feathering the calibrated amblyopic eye region. In practice, firstly, for each boundary pixel in the aforementioned set of boundary pixels, the brightness level configuration server uses a Gaussian blur algorithm to determine a blur window with a preset Gaussian radius centered on the boundary pixel. Then, a weighted average is calculated on the brightness values ​​of all pixels within the blur window, and the result of the weighted average calculation is used as the updated brightness value of the boundary pixel. Then, the regions whose brightness values ​​of all boundary pixels in the aforementioned modified calibration contralateral healthy eye region and the aforementioned calibrated amblyopic eye region are updated are respectively defined as the feathered calibration contralateral healthy eye region and the feathered calibration amblyopic eye region. It should be noted that the pixels at the center of the aforementioned blurred window have the highest weight, and the weight gradually decreases towards the edge of the blurred window. For example, the preset Gaussian radius can be 2 pixels.

[0030] The sixth sub-step involves determining the calibration search duration for the target object based on the feathered calibration of the contralateral healthy eye region, the feathered calibration of the amblyopic eye region, and a preset duration threshold. The calibration search duration represents the time taken from the start of the search to finding the target object. The preset duration threshold represents a pre-set threshold, such as 10 seconds. In practice, firstly, the brightness level configuration server can display the feathered calibration of the contralateral healthy eye region and the feathered calibration of the amblyopic eye region to the target object via the user terminal device. Then, a timer is used to obtain the time taken from the start of the search to finding the target object as the initial search duration. Next, in response to determining that the initial search duration is less than the preset duration threshold, the initial search duration is determined as the calibration search duration. In response to determining that the initial search duration is greater than or equal to the preset duration threshold, the preset duration threshold is determined as the calibration search duration. It should be noted that the search start time mentioned above is the time when the aforementioned user terminal device begins to display the feathered calibration of the contralateral healthy eye region and the feathered calibration of the amblyopic eye region.

[0031] Then, the average calibration search time is calculated by averaging the obtained calibration search times. This average calibration search time represents the average of the individual calibration search times. In practice, the brightness level configuration server can use an arithmetic average to calculate the average calibration search time.

[0032] The second step involves determining the training brightness level based on the obtained average calibration search times, and then performing amblyopia training based on these training brightness levels. In practice, the brightness level configuration server first uses a comparison and sorting algorithm to sort the obtained average calibration search times, resulting in sorted average calibration search times. Then, the preset brightness level corresponding to the smallest average calibration search time among the sorted average calibration search times is determined as the training brightness level. It should be noted that the specific implementation steps for amblyopia training based on the above training brightness levels can be found in step 203, and will not be elaborated upon here.

[0033] The above-described technical solution, as an inventive point of this disclosure, addresses technical problem two: "The excessive computational and storage resources wasted in amblyopia training result in high server load, long overall system training time, and poor training effectiveness." The reasons for this are as follows: Significant differences in binocular inhibition levels among different users mean that a fixed training brightness level cannot meet individual needs. This leads to frequent timeouts for some users due to excessively high brightness levels, or ineffective training due to excessively low brightness levels. The system needs to repeatedly execute invalid training sessions, recording a large amount of failure data, which not only wastes significant computational and storage resources but also greatly increases server load, resulting in long overall system training time and poor training effectiveness. Solving these factors reduces the wasted computational and storage resources in amblyopia training, lowers server load, shortens overall system training time, and improves training effectiveness. To achieve this effect, the target-search-based amblyopia training system disclosed herein, through a brightness level configuration server, can iterate through each calibration image corresponding to each brightness level to determine the calibration search time for the user to find the target item, based on a preset set of brightness levels and a set of calibration images. Based on the obtained calibration search times, the average search time corresponding to each brightness level is calculated. Different brightness levels are then selected, and the brightness level with the shortest average search time is used as the training brightness level, fundamentally solving the problem of level mismatch. This reduces invalid training sessions and the recording of failed data, thereby reducing the wasted computational and storage resources for amblyopia training, lowering server load, shortening the overall system training time, and improving training effectiveness.

[0034] Step 202: The personalized image generation server generates a set of test images based on the user account information and the set of background materials.

[0035] In some embodiments, the personalized image generation server can generate a set of test images based on user account information and a set of background materials. The user account information can represent a unique identifier corresponding to the target object. The background materials in the set of background materials can represent pre-constructed scene images corresponding to the weighted items. The scene images can represent images displaying a specific scene. Here, the specific type of the scene is not limited; for example, the scene could be a restaurant. The test images in the set of test images can represent images containing the target item used for performing amblyopia training operations or healthy eye training operations.

[0036] In addressing the technical problems mentioned above, and considering the application scenario of amblyopia training for preschool and school-aged children with amblyopia (e.g., children aged 3-12), the following technical problem often arises: Significant individual differences exist among users in terms of age, gender, and interests. Using generic test images for training leads to a lack of motivation for users to actively search for uninteresting test images, resulting in shorter fixation time and decreased search intent. Furthermore, the training content is monotonous and repetitive, easily causing user boredom and leading to a large number of users dropping out prematurely, wasting generated training images and allocated computing resources. Therefore, considering the following requirements for this application scenario: adapting to age differences, gender preferences, and changing interests of different users, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the personalized image generation server described above can generate a set of test images based on user account information and a set of background materials through the following steps: The first step is to obtain objective physiological characteristics and subjective interest characteristics based on the aforementioned user account information. The objective physiological characteristics can represent the biological attributes of the target object, such as age. The subjective interest characteristics can represent a list of the target object's preferences for various target items, ranked from highest to lowest. Each entry in the list can include a target item and its corresponding ranking. The specific category of each target item is not limited; for example, a target item could be a white car. The ranking represents the position of the target object's preference for each target item from highest to lowest. In practice, the personalized image generation server can use a query statement in the database, using the user account information as a query index, to retrieve the objective physiological characteristics and subjective interest characteristics from a pre-stored user database. The pre-stored user database represents a pre-stored database recording the correspondence between the user account information, the objective physiological characteristics, and the subjective interest characteristics. For example, the user database stores the following record: "User account information: 10001; Objective physiological characteristics (age): 10; Subjective interest characteristics: [...{Target item: white car; Item sorting position: 1}...]".

[0037] The second step involves generating a weighted item set based on the ranking of the aforementioned subjective interest features. This weighted item set may include a preset number of weighted items. Each weighted item in the set represents a target item generated based on the aforementioned subjective interest features for generating test images one by one. In practice, the personalized image generation server first uses a roulette wheel selection method to assign a probability of occurrence to each target item according to its ranking, with items having lower rankings receiving higher probabilities. Then, the preset number is multiplied by the probability corresponding to each target item to obtain the frequency of each target item's occurrence. Finally, based on the frequency of occurrence of each target item, each target item is stored in a list, resulting in a set containing the preset number of target items as the weighted item set. For example, if the target item is a white car, its probability is 0.5, and the preset number is 60, then the target item will appear 30 times, and the weighted item set will contain 30 white cars.

[0038] Third, based on each weighted item in the above set of weighted items, perform the following image generation steps: The first sub-step involves generating background materials corresponding to the weighted items as background style templates based on the aforementioned objective physiological characteristics and the aforementioned set of background materials. These background style templates can represent the scene image used as the underlying layer of the test image; for example, a background style template could be an image of a child's room. In practice, the personalized image generation server can use a tag matching algorithm, with the aforementioned objective physiological characteristics as query conditions, to retrieve the background material set and obtain the background materials corresponding to the weighted items as background style templates. It should be noted that there is a pre-defined correspondence between the background materials in the aforementioned set of background materials and the aforementioned objective physiological characteristics. This correspondence is stored in the form of a mapping table. This correspondence can include the value (or value range) of the objective physiological characteristics, or the correspondence between the category of the objective physiological characteristics and the aforementioned background materials. For example, the correspondence can include, but is not limited to: "Objective physiological characteristics (age): 7 years old (or 6-8 years old range); Background material: Image of a child's room", or "Objective physiological characteristics (age): 25 years old; Background material: Image of a simple living room", or "Objective physiological characteristics (gender): Female; Background material: Image of flowers everywhere".

[0039] The second sub-step involves generating a bounding rectangular region corresponding to the weighted object based on its contour features. The contour features characterize the boundary shape of the weighted object. The bounding rectangular region represents the smallest rectangle that can completely enclose the weighted object. In practice, the personalized image generation server can use a contour tracking algorithm to scan the weighted object and obtain the bounding rectangular region.

[0040] The third sub-step involves determining the placement position of the aforementioned circumscribed rectangular region based on fixed coordinates. These fixed coordinates can represent pre-defined two-dimensional coordinates. For example, the fixed coordinates could be (780, 30). The placement position represents the coordinates of the circumscribed rectangular region within the aforementioned background style template. In practice, the personalized image generation server can use coordinate assignment operations to set the top-left corner coordinates of the circumscribed rectangular region to the aforementioned fixed coordinates to obtain the placement position.

[0041] The fourth sub-step involves pre-clearing the pixels in the background style template corresponding to the bounding rectangle region based on the aforementioned placement position, resulting in a reserved compositing area. This reserved compositing area represents the blank area left after the pre-clearing process. In practice, the personalized image generation server first calculates the range of pixels to be cleared in the background style template using a memory clearing operation, combined with the placement position and the size of the bounding rectangle region. Then, the color values ​​of all pixels within this range are set to zero, resulting in the reserved compositing area.

[0042] The fifth sub-step involves performing anisotropic blurring on the edge pixels of the aforementioned weighted item to obtain an edge-blurred weighted item. Here, the aforementioned edge pixels represent the pixels on the outline of the aforementioned weighted item. The aforementioned edge-blurred weighted item represents the item obtained after anisotropic blurring of the aforementioned weighted item. The aforementioned anisotropic blurring process represents an image smoothing operation using different blur radii in the horizontal and vertical directions. In practice, the aforementioned personalized image generation server can use anisotropic Gaussian blurring to blur the edge pixels of the aforementioned weighted item to obtain an edge-blurred weighted item. It should be noted that the blur radius in the horizontal direction is larger than the blur radius in the vertical direction.

[0043] The sixth sub-step involves generating a complete original image based on the aforementioned edge-blurred weighted items and the aforementioned background style template. This complete original image can represent a full-color image containing the aforementioned target items and the aforementioned background style template, without view splitting. In practice, the aforementioned personalized image generation server can use image overlay operations to overlay the aforementioned edge-blurred weighted items onto the aforementioned reserved composite area to obtain the complete original image.

[0044] The seventh sub-step involves performing composite interference processing on the aforementioned complete original image based on the initial difficulty level, resulting in the final complete image. Here, the initial difficulty level can represent the training difficulty level at which the amblyopia training operation begins. The initial difficulty level can represent any training difficulty level in a preset difficulty gradient; for example, the initial difficulty level could be the first level of the second difficulty. The composite interference processing can represent the image processing operation of simultaneously adding different numbers of interference objects and superimposing different numbers of gratings. The final complete image can represent the complete original image after the addition of interference objects and the superposition of gratings. In practice, firstly, the personalized image generation server can generate interference objects in the edge region of the aforementioned complete original image using uniform sampling, according to the number of interference objects corresponding to the initial difficulty level. Then, using bilinear interpolation, gratings are superimposed on the weighted objects and each interference object according to the number of gratings corresponding to the initial difficulty level, to obtain the final complete image.

[0045] The fourth step is to determine the final complete images obtained as the test image set.

[0046] The above-described technical solution, as an inventive point of this disclosure, addresses technical problem three: "Amblyopia training consumes a large amount of computational and storage resources." The reasons for this waste of computational and storage resources in amblyopia training are as follows: Significant individual differences exist among users in terms of age, gender, and interests. Using generic test images for training leads to a lack of motivation for users to actively search for uninteresting test images, resulting in shorter fixation time and decreased search intent. Furthermore, the training content is monotonous and repetitive, easily causing user boredom and leading to many users dropping out prematurely, wasting generated training images and allocated computational resources. Solving these factors can reduce the waste of computational and storage resources in amblyopia training. To achieve this effect, the target search-based amblyopia training system disclosed herein utilizes a personalized image generation server. This server acquires objective physiological characteristics based on the user's age and gender from their account information and generates a weighted set of items according to their interest ranking. This ensures that target items with higher rankings appear more frequently in the test image set. Simultaneously, a corresponding background style template is generated for each weighted item, and edge blurring is applied. Ultimately, a test image set highly matched to the user's interests and preferences is generated. This maintains the user's high motivation to search for test images, reducing the waste of training images and computational resources due to user withdrawals. Therefore, it reduces the computational and storage resources wasted on amblyopia training, improving training effectiveness.

[0047] Step 203: The amblyopia training server performs the following amblyopia training steps based on each test image in the above test image set and the training difficulty level: Step 2031: Perform irregular segmentation on the test image to obtain the contralateral healthy eye region and the amblyopic eye region.

[0048] In some embodiments, the amblyopia training server can perform irregular segmentation processing on the test image to obtain the contralateral healthy eye region and the amblyopic eye region. The training difficulty level can characterize the task complexity during amblyopia training. The training difficulty level can be either a first difficulty or a second difficulty. The first difficulty can characterize the proportion of the rectangular frame containing the target object to the entire test image as a first preset proportion. The first preset proportion can be 4%. The second difficulty can characterize the proportion of the rectangular frame containing the target object to the entire test image as a second preset proportion. The second preset proportion can be 1%. Both the first and second difficulties can include five levels. These five levels can be divided according to the number of interfering objects: level one corresponds to one interfering object, level two to two, level three to three, level four to four, and level five to five. The interfering objects can represent objects obtained by superimposing a preset number of gratings on top of the target object with the same appearance. The specific value of the preset number is not limited and can be set according to actual needs. The five levels of the first difficulty and the five levels of the second difficulty mentioned above are combined in ascending order of training difficulty to form a preset level gradient. The preset level gradient, from lowest to highest, is as follows: Level 1 of the first difficulty, Level 2 of the first difficulty, Level 3 of the first difficulty, Level 4 of the first difficulty, Level 5 of the first difficulty, Level 1 of the second difficulty, Level 2 of the second difficulty, Level 3 of the second difficulty, Level 4 of the second difficulty, and Level 5 of the second difficulty. The aforementioned amblyopia training can represent the process of training the amblyopic eye of the target object using the aforementioned set of test images. The aforementioned healthy eye training can represent the process of training the contralateral healthy eye of the target object using the aforementioned set of test images. The aforementioned contralateral healthy eye region can represent the visual field range allocated to the contralateral healthy eye of the target object after irregular segmentation of the aforementioned test images. The aforementioned amblyopia region can represent the visual field range allocated to the amblyopic eye of the target object after irregular segmentation of the aforementioned test images. In practice, the aforementioned amblyopia training server can use a threshold segmentation algorithm to irregularly segment the test image, obtaining the contralateral healthy eye region and the amblyopic eye region corresponding to the test image. It should be noted that the area of ​​the contralateral healthy eye region and the area of ​​the amblyopic eye region each account for 50% of the area of ​​the test image.

[0049] Step 2032: Based on the training brightness level, modify the contralateral healthy eye area to obtain the training contralateral healthy eye area.

[0050] In some embodiments, the amblyopia training server can modify the contralateral healthy eye region according to the aforementioned training brightness level to obtain a training contralateral healthy eye region. The training contralateral healthy eye region can represent the visual field range obtained after adjusting the brightness level of the contralateral healthy eye region to the aforementioned training brightness level. In practice, the amblyopia training server can use a linear transformation algorithm to adjust the brightness of the contralateral healthy eye region to the aforementioned training brightness level to obtain the training contralateral healthy eye region.

[0051] Step 2033: Based on the training contralateral healthy eye area and amblyopic eye area, control the user terminal device to perform amblyopic eye training operations and obtain amblyopic eye training results.

[0052] In some embodiments, the amblyopia training server can control the user terminal device to perform amblyopia training operations based on the contralateral healthy eye region and the amblyopia region, thereby obtaining amblyopia training results.

[0053] In some optional implementations of certain embodiments, the aforementioned amblyopia training server may be further configured to perform the following steps to control the user terminal device to perform amblyopia training operations based on the contralateral healthy eye region and the amblyopia region, thereby obtaining amblyopia training results: The first step, based on the aforementioned training of the contralateral healthy eye region, the aforementioned amblyopic eye region, and the aforementioned preset duration threshold, is to perform the following training steps: First, the search duration and eye movement data of the target object are acquired. The search duration represents the time taken from the start of the search to finding the target object in the amblyopic region. The eye movement data represents the fixation point position, fixation duration, and fixation trajectory of the target object. The fixation point position represents the coordinates of the fixation point. The fixation duration represents the duration of fixation in the amblyopic region. The fixation trajectory represents the sequence of coordinate points along the fixation point's movement. In practice, the amblyopia training server can use a counter to acquire the time taken from the start of the search to finding the target object in the amblyopic region as the search duration, and acquire the eye movement data of the target object using an eye-tracking device. For example, the eye-tracking device could be a Tobii Pro Spectrum, or a far-viewing screen with eye tracking.

[0054] Then, in response to determining that the search duration is less than the preset duration threshold, an amblyopia training result is generated based on the search duration and the test image. The amblyopia training result can characterize the result of amblyopia training on the target object. The amblyopia training result can include gaze type, training search duration, and test image. The gaze type can characterize the target object's gaze on the target object. The gaze type can characterize a first gaze type, a second gaze type, or a third gaze type. The first gaze type can characterize that the target object gazed at the target object and successfully found it. The second gaze type can characterize that the target object gazed at the target object but failed to find it. The third gaze type can characterize that the target object did not gaze at the target object. The training search duration can characterize the time spent by the target object in one trial from the start of the search to finding the target object. One trial can characterize performing an amblyopia training operation on a test image. In practice, firstly, the amblyopia training server can determine the search duration as the training search duration. Then, the first gaze type is determined as the gaze type of the target object. Finally, the training search time, the gaze type, and the test image are determined as the training results for amblyopia.

[0055] Finally, in response to determining that the search duration equals the preset duration threshold, amblyopia training results are generated based on the preset duration threshold, the eye-tracking data, and the test image. It should be noted that the longest search duration set in a single trial does not exceed the preset duration threshold; therefore, cases where the search duration exceeds the preset duration threshold are not specifically analyzed.

[0056] In some optional implementations of certain embodiments, in response to determining that the search duration is equal to the preset duration threshold, the amblyopia training server may be further configured to perform the following steps to generate amblyopia eye training results based on the preset duration threshold, the eye-tracking data, and the test image: The first step is to determine the above-mentioned preset time threshold as the training search time.

[0057] The second step is to generate a gaze type based on the aforementioned eye-tracking data. In practice, firstly, the aforementioned amblyopia training server can extract and process the aforementioned eye-tracking data using the aforementioned eye-tracking device to obtain the gaze coordinates corresponding to the aforementioned target object. The gaze coordinates can represent the coordinates of the gaze point of the aforementioned target object in the aforementioned amblyopic eye region. Then, using the PtInRect function, the gaze coordinates are compared with the coordinate range of the aforementioned amblyopic eye region. In response to detecting that the gaze coordinates are within the coordinate range of the aforementioned amblyopic eye region, the aforementioned second gaze type is determined as the gaze type of the aforementioned target object. In response to detecting that the gaze coordinates are not within the coordinate range of the aforementioned amblyopic eye region, the aforementioned third gaze type is determined as the gaze type of the aforementioned target object.

[0058] The third step is to determine the above-mentioned training search duration, above-mentioned gaze type, and above-mentioned test image as the training results for amblyopia.

[0059] Step 2034: Based on the training results for amblyopia, the training difficulty level is modified to obtain the updated training difficulty level.

[0060] In some embodiments, the amblyopia training server can modify the training difficulty level based on the amblyopia training results to obtain an updated training difficulty level. The updated training difficulty level represents the training difficulty level obtained after modifying the original training difficulty level.

[0061] The first step involves upgrading the training difficulty level in response to the detection that the fixation type included in the amblyopia training results is the first fixation type, thereby obtaining an upgraded training difficulty level. In practice, the amblyopia training server can increase the training difficulty level by one level according to the preset level gradient to obtain the upgraded training difficulty level.

[0062] The second step involves downgrading the training difficulty level in response to the detection that the fixation type included in the amblyopia training results is either the second or third fixation type. In practice, the amblyopia training server can reduce the training difficulty level by one level according to the preset level gradient to obtain the downgraded training difficulty level.

[0063] Third, in response to the detection that the training search time exceeds the preset time threshold, the training difficulty level is downgraded to obtain a downgraded training difficulty level. In practice, the aforementioned amblyopia training server can reduce the training difficulty level by one level according to the preset level gradient to obtain a downgraded training difficulty level.

[0064] Step 2035: Use the updated training difficulty level as the training difficulty level for the next test image and continue with the amblyopia training steps.

[0065] In some embodiments, the amblyopia training server can use the updated training difficulty level as the training difficulty level for the next test image and continue executing the amblyopia training steps. The next test image can represent the next image used to perform the amblyopia training operation.

[0066] Step 2036: In response to the detection that the number of times the amblyopia training step has been executed meets the preset number condition, terminate the amblyopia training and obtain the training results for each amblyopia.

[0067] In some embodiments, in response to detecting that the number of times the above-mentioned amblyopia training steps are executed meets a preset number condition, the amblyopia training server can terminate the above-mentioned amblyopia training and obtain various amblyopia training results. The above-mentioned number of executions can represent the number of times the above-mentioned amblyopia training operation is performed. The above-mentioned preset number condition can be that the number of executions equals a preset number threshold. The above-mentioned preset number threshold can be 60.

[0068] Optionally, after step 203 above, the aforementioned amblyopia training server may also perform the following steps: The first step, in response to determining that the number of amblyopia training sessions meets the preset ratio condition, is to perform the following steps: First, based on the aforementioned training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. The preset ratio can be a 12:1 ratio between the number of times the amblyopia training operation is performed and the number of times the healthy eye training operation is performed. The training contralateral healthy eye region represents the visual field obtained after adjusting the brightness level of the contralateral healthy eye region to the aforementioned training brightness level. In practice, the amblyopia training server can use a linear transformation algorithm to adjust the brightness of the target contralateral healthy eye region to the aforementioned training brightness level to obtain the training contralateral healthy eye region.

[0069] Then, based on the aforementioned amblyopic eye region and the aforementioned contralateral healthy eye region, the user terminal device is controlled to perform an eye-training operation to obtain the eye-training result. This eye-training result characterizes the outcome of eye-training on the target object. The eye-training result may include fixation type, training search duration, and test image. It should be noted that the method of controlling the user terminal device to perform the eye-training operation based on the aforementioned amblyopic eye region and the aforementioned contralateral healthy eye region to obtain the eye-training result is the same as the method of controlling the user terminal device to perform the amblyopic eye training operation based on the aforementioned contralateral healthy eye region and the aforementioned amblyopic eye region to obtain the amblyopic eye training result; refer to step 2033, which will not be repeated here. Figure 5 As shown, Figure 5 This is a physical diagram of the internal test device used to perform the above-mentioned amblyopia training operation or the above-mentioned healthy eye training operation. Figure 5 The user-end device shown may include internal structures such as a display module, optical lenses, and sensor components. The display module can be used to present the test image in the contralateral healthy eye area and amblyopic eye area to perform the amblyopic eye training operation or the healthy eye training operation. The sensor components can collect feedback data from the user during the amblyopic eye training operation or the healthy eye training operation, and send the feedback data to the amblyopic eye training server, so that the amblyopic eye training server can generate amblyopic eye training results or healthy eye training results based on the feedback data. The feedback data may include, but is not limited to, the user's search time for the target item and the user's eye movement data. For example, the display module may be an LCD liquid crystal display, the optical lens may be a freeform optical lens, and the sensor component may be an eye-tracking sensor (Tobii Pro Spectrum).

[0070] In addressing the technical problems mentioned above, and considering the application scenario of training amblyopia in children with small-angle strabismus and unstable foveal fixation, the following technical problem often arises: These users exhibit poor fixation stability and their fixation points are prone to drift. They frequently fail to fixate on the target due to insufficient fixation time, or they accidentally click on the target without fixation and are judged as successful. Existing systems adjust training difficulty solely based on whether the user finds the target, failing to differentiate between these situations. This leads to frequent errors in training difficulty adjustment, generating a large number of invalid training sessions and failure data. The system needs to process more training sessions, significantly increasing server load and overall training time. To address the following requirements for this application scenario: adapting to differences in fixation stability and anti-interference capabilities among users, we have decided to adopt the following solution: Following step 203 above, the aforementioned amblyopia training server is further configured to update the training difficulty level for the next round through the following steps: The first step, based on each amblyopia training result from the above-mentioned amblyopia training results, is to perform the following steps: The first sub-step involves generating user eye-tracking data based on the acquired user eye images. The user eye images represent images of the target object's eye region captured by a camera. The user eye-tracking data represents parameters of the target object's eye movements. The user eye-tracking data may include a gaze point coordinate sequence and a gaze time set. Each gaze point coordinate in the gaze point coordinate sequence represents the two-dimensional coordinates of the target object's gaze point on the user's device. The gaze time in the gaze time set represents the duration of gaze at each gaze point in the gaze point coordinate sequence. In practice, firstly, the amblyopia training server uses an eye-tracking algorithm to extract and process the user eye images to obtain a gaze point coordinate sequence, simultaneously recording the start and end times of each gaze point. The difference between each start time and its corresponding end time is determined as the gaze time set. Then, the gaze point coordinate sequence and the gaze time set are used to define the user eye-tracking data. The start time represents the moment when the gaze point begins to gaze. The end time represents the moment when the gaze point ends to gaze. The aforementioned next round of training difficulty level can represent the training difficulty level corresponding to the next amblyopia training operation or the healthy eye training operation.

[0071] The second sub-step generates the target object fixation duration, distractor fixation duration, and total fixation duration based on the aforementioned user eye-tracking data. The target object fixation duration represents the duration the target object is fixated on. The distractor fixation duration represents the duration the target object is fixated on. The total fixation duration represents the sum of fixation times for all fixation points on the test image. In practice, the amblyopia training server can use a ray casting method to traverse the coordinates of each fixation point in the fixation point coordinate sequence, determining the sum of fixation times for each fixation point within the target object region as the target object fixation duration, the sum of fixation times for each fixation point within the distractor object region as the distractor object fixation duration, and the sum of fixation times for all fixation points corresponding to the test image as the total fixation duration.

[0072] The third sub-step involves determining a first ratio and a second ratio based on the fixation duration of the target object, the total fixation duration, and the fixation duration of the interfering object. The first ratio represents the ratio of the fixation duration of the target object to the total fixation duration. The second ratio represents the ratio of the fixation duration of the interfering object to the total fixation duration. In practice, the amblyopia training server can use division to obtain the first ratio by dividing the fixation duration of the target object by the total fixation duration, and the second ratio by dividing the fixation duration of the interfering object by the total fixation duration.

[0073] The fourth sub-step involves calculating the spatial distribution entropy of the aforementioned fixation point coordinate sequence to obtain the fixation entropy. The fixation entropy characterizes the degree of disorder in the spatial distribution of fixations; a higher entropy value indicates a more dispersed fixation point distribution, while a lower entropy value indicates a more concentrated fixation point distribution. In practice, firstly, the aforementioned amblyopia training server can divide the screen of the user terminal device into a preset number of grids using a grid partitioning method, with each grid corresponding to an index. The specific value of this preset number is not limited and can be set according to actual needs. Then, the number of fixation point coordinates in the aforementioned fixation point coordinate sequence is determined as the first value. Next, for each grid in the preset number of grids, the following steps are performed: First, traverse each fixation point coordinate in the aforementioned fixation point coordinate sequence, count the number of fixation point coordinates falling within the grid, and obtain the count value for each grid as the second value. Then, the ratio of the second value to the first value is determined as the probability value of the grid. Finally, the obtained probability values ​​are input into the Shannon entropy calculation formula to obtain the fixation entropy. As an example, the Shannon entropy calculation formula can be as follows: .

[0074] Among them, the above This can characterize the obtained gaze entropy. The above This can represent the number of grids obtained after dividing the screen of the aforementioned user terminal device. This can represent the index corresponding to each grid in the aforementioned preset number of grids. The above... This can represent the index of the above-preset number of grids. The probability value corresponding to the grid.

[0075] The fifth sub-step involves fusing the first ratio, the second ratio, and the gaze entropy to obtain a visual search efficiency coefficient. This visual search efficiency coefficient is a quantitative indicator that describes the efficiency of searching for the target item using the target object. In practice, the amblyopia training server can use a linear weighted summation method to obtain the visual search efficiency coefficient by weighting and summing the first ratio, the second ratio, and the gaze entropy.

[0076] The second step is to average the obtained visual search efficiency coefficients to obtain the average visual search efficiency. This average visual search efficiency represents the average of the individual visual search efficiency coefficients. In practice, the aforementioned amblyopia training server can use an arithmetic mean to average the obtained visual search efficiency coefficients to obtain the average visual search efficiency.

[0077] The third step involves downgrading the current difficulty level in response to the detection that the average visual search efficiency meets the preset retrieval criteria. The preset retrieval criteria indicate that the average visual search efficiency is less than a preset efficiency threshold. The preset efficiency threshold represents a pre-defined value used to determine whether the average visual search efficiency meets the standard. The specific value of the average visual search efficiency is not limited and can be set according to actual needs. The current difficulty level represents the training difficulty level of the current training round. The downgraded difficulty level represents the difficulty level obtained after downgrading the current difficulty level. In practice, the amblyopia training server can reduce the current difficulty level by one level according to the preset level gradient to obtain the downgraded difficulty level.

[0078] Fourth, in response to the detection that the average visual search efficiency does not meet the preset retrieval conditions, the current difficulty level is upgraded to obtain an upgraded difficulty level. The upgraded difficulty level represents the difficulty level obtained after upgrading the current difficulty level. In practice, the amblyopia training server can increase the current difficulty level by one level according to the preset level gradient to obtain the upgraded difficulty level.

[0079] Fifth, use either the downgraded or upgraded difficulty level mentioned above as the training difficulty level to execute the next round of amblyopia training. It should be noted that one round of amblyopia training may include a preset number of trials. This preset number can be 60. Each amblyopia training result corresponds to one of these trials.

[0080] The above technical solution, as an inventive point of this disclosure, solves technical problem four: "leading to a significant increase in server load and a long overall system training time." The reasons for this significant increase in server load and long overall system training time are as follows: These users have poor fixation stability and their fixation points are prone to drift. They frequently experience situations where they fixate on a target but are judged as failing due to insufficient fixation time, or they accidentally click and are judged as successful without fixing on the target. Existing systems only adjust training difficulty based on whether the user finds the target, failing to distinguish between these situations. This leads to frequent errors in training difficulty adjustment, generating a large number of invalid training sessions and failure data. The system needs to process more training sessions, resulting in a significant increase in server load and a long overall system training time. Solving these factors can reduce server load, shorten the overall system training time, and improve training effectiveness. To achieve this effect, the target-search-based amblyopia training system of this disclosure, through an amblyopia training server, can generate the fixation time of the target object, the fixation time of the distractor, and the total fixation time based on eye-tracking data, and calculate a first ratio and a second ratio. The gaze entropy is calculated by performing spatial distribution entropy calculations on the gaze point coordinate sequence. The first ratio, second ratio, and gaze entropy are then fused to obtain the visual search efficiency coefficient. The visual search efficiency coefficient for each trial is averaged and compared with a preset efficiency threshold. Based on the comparison results, the training difficulty level is dynamically adjusted. This allows for personalized adjustment of training difficulty based on the user's actual gaze behavior quality, reducing invalid training sessions and failed data caused by incorrect difficulty adjustment. Consequently, server load is reduced, overall system training time is shortened, and training effectiveness is improved.

[0081] The above-described embodiments of this disclosure have the following beneficial effects: A target-search-based amblyopia training system according to some embodiments of this disclosure can reduce the computational and storage resources consumed in amblyopia training, lower server load, shorten the overall system training time, and improve the effectiveness of amblyopia training. Specifically, the reasons for the high computational and storage resources consumed in amblyopia training, high server load, long overall system training time, and poor training effect are as follows: Existing systems typically use fixed training parameters (such as fixed occlusion time and fixed training content), failing to consider individual differences among users in terms of age, gender, and interests, resulting in inconsistent training effects. Some users require longer training periods to achieve visual function improvement. The system needs to generate and store more training images and record more training data, which not only wastes a large amount of computational and storage resources but also significantly increases server load, leading to a long overall system training time and poor training effect. Based on this, some embodiments of the target search-based amblyopia training system disclosed herein include: a brightness level configuration server, a personalized image generation server, an amblyopia training server, and a user terminal device that are interconnected. The brightness level configuration server is used to determine the training brightness level based on a preset brightness level set and an acquired calibration image set. Thus, the brightness level for amblyopia training can be obtained. The personalized image generation server is used to generate a test image set based on user account information and a background material set. Thus, a test image set for amblyopia training can be obtained. The amblyopia training server is used to perform the following amblyopia training steps based on each test image in the test image set and the training difficulty level: First, the test images are irregularly segmented to obtain the contralateral healthy eye region and the amblyopia region. Thus, the segmented contralateral healthy eye region and the amblyopia region can be obtained. Then, based on the training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. Thus, the training contralateral healthy eye region for amblyopia training can be obtained. Next, based on the training regions of the contralateral healthy eye and the amblyopic eye, the user terminal device is controlled to perform amblyopic eye training operations to obtain amblyopic eye training results. This allows control over the user terminal device to perform amblyopic eye training operations. Then, based on the amblyopic eye training results, the training difficulty level is modified to obtain an updated training difficulty level. This updated training difficulty level is then used as the training difficulty level for the next test image, and the amblyopic eye training steps continue. Finally, in response to the detection that the number of executions of the amblyopic eye training steps meets a preset condition, the amblyopic eye training is terminated, and each amblyopic eye training result is obtained. This completes one round of amblyopic eye training.Because it doesn't use fixed training parameters, but instead determines personalized training brightness levels through a brightness level configuration server, the training brightness levels can adapt to the binocular inhibition levels of different users, reducing the number of invalid training sessions caused by brightness level mismatches, thereby reducing the server's computational load. Also, because it generates personalized test image sets through a personalized image generation server and performs amblyopia training based on these personalized test image sets, it reduces the waste of generated training images and allocated computing resources caused by users quitting midway due to boredom. Furthermore, because the amblyopia training server dynamically adjusts the training difficulty level based on the gaze type in the amblyopia training results, it adjusts the training difficulty level based on the user's actual gaze behavior quality, reducing repetitive and invalid training caused by inaccurate difficulty adjustments. This reduces the number of training images and training data that the system needs to generate and store, thus shortening the overall system training time and improving the amblyopia training effect.

[0082] Continue to refer to Figure 3 , Figure 3 A flow 300 of some embodiments of the object search-based weak vision training method according to this disclosure is shown. The object search-based weak vision training method includes the following steps: Step 301: Determine the training brightness level based on the preset brightness level set and the acquired calibration image set.

[0083] In some embodiments, the execution subject (e.g., a computing device) of the target search-based amblyopia training method can determine the training brightness level based on a preset brightness level set and the acquired calibration image set.

[0084] Step 302: Generate a set of test images based on user account information and background material set.

[0085] In some embodiments, the aforementioned execution entity may generate a set of test images based on user account information and a set of background materials.

[0086] Step 303: Based on each test image in the test image set and the training difficulty level, perform the following amblyopia training steps: Step 3031: Perform irregular segmentation on the test image to obtain the contralateral healthy eye region and the amblyopic eye region.

[0087] In some embodiments, the execution entity may perform irregular segmentation processing on the test image to obtain the contralateral healthy eye region and the amblyopic eye region.

[0088] Step 3032: Based on the training brightness level, modify the contralateral healthy eye area to obtain the training contralateral healthy eye area.

[0089] In some embodiments, the execution entity may modify the contralateral healthy eye region according to the training brightness level to obtain the training contralateral healthy eye region.

[0090] Step 3033: Based on the training contralateral healthy eye area and amblyopic eye area, control the user terminal device to perform amblyopic eye training and obtain the amblyopic eye training result.

[0091] In some embodiments, the execution entity can control the user terminal device to perform amblyopia training operations based on the contralateral healthy eye region and the amblyopia region, thereby obtaining amblyopia training results.

[0092] Step 3034: Based on the training results for amblyopia, the training difficulty level is modified to obtain the updated training difficulty level.

[0093] In some embodiments, the execution entity may modify the training difficulty level based on the amblyopia training results to obtain an updated training difficulty level.

[0094] Step 3035: Use the updated training difficulty level as the training difficulty level for the next test image and continue with the amblyopia training steps.

[0095] In some embodiments, the execution entity may use the updated training difficulty level as the training difficulty level for the next test image and continue to execute the amblyopia training steps.

[0096] Step 3036: In response to the detection that the number of times the amblyopia training step has been executed meets the preset number condition, the amblyopia training is terminated, and the training results for each amblyopia are obtained.

[0097] In some embodiments, in response to detecting that the number of times the above-mentioned amblyopia training steps are executed meets a preset number condition, the execution entity may terminate the above-mentioned amblyopia training and obtain each amblyopia training result.

[0098] The following is for reference. Figure 4 It shows a schematic diagram of the structure of an electronic device 400 (e.g., a computing device) suitable for implementing some embodiments of the present disclosure. Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0099] like Figure 4As shown, electronic device 400 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 402 or a program loaded from storage device 408 into random access memory (RAM) 403. RAM 403 also stores various programs and data required for the operation of electronic device 400. Processing device 401, ROM 402, and RAM 403 are interconnected via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0100] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic device 400 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 An electronic device 400 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 4 Each box shown can represent a device or multiple devices as needed.

[0101] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 409, or installed from storage device 408, or installed from ROM 402. When the computer program is executed by processing device 401, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0102] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0103] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0104] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the steps between steps 301 and 3037.

[0105] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0107] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0108] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A weak vision training system based on target search, wherein, The amblyopia training system includes: a brightness level configuration server, a personalized image generation server, an amblyopia training server, and user terminal devices that are interconnected. The brightness level configuration server is used to determine the training brightness level based on the preset brightness level set and the acquired calibration image set. The personalized image generation server is used to generate a set of test images based on user account information and a set of background materials; The amblyopia training server is configured to perform the following amblyopia training steps based on each test image in the test image set and the training difficulty level: irregularly segmenting the test image to obtain a contralateral healthy eye region and an amblyopia region; modifying the contralateral healthy eye region according to the training brightness level to obtain a training contralateral healthy eye region; controlling the user terminal device to perform amblyopia training operations based on the training contralateral healthy eye region and the amblyopia region to obtain an amblyopia training result; modifying the training difficulty level based on the amblyopia training result to obtain an updated training difficulty level; using the updated training difficulty level as the training difficulty level for the next test image and continuing to execute the amblyopia training steps; and terminating the amblyopia training in response to detecting that the number of executions of the amblyopia training steps meets a preset number condition, thereby obtaining each amblyopia training result.

2. The system according to claim 1, wherein, The amblyopia training server is configured as follows: Based on the training contralateral healthy eye region, the amblyopic eye region, and the preset duration threshold, the following steps are performed: Acquire search duration and eye-tracking data for the target item; In response to determining that the search duration is less than the preset duration threshold, amblyopia training results are generated based on the search duration and the test image; In response to determining that the search duration is equal to the preset duration threshold, amblyopia training results are generated based on the preset duration threshold, the eye-tracking data, and the test image.

3. The system according to claim 2, wherein, The amblyopia training server is configured as follows: In response to determining that the search duration is equal to the preset duration threshold, the preset duration threshold is determined as the training search duration; Based on the eye movement data, a gaze type is generated; The training search duration, the gaze type, and the test image are determined as the training results for amblyopia.

4. The system according to claim 1, wherein, The amblyopia training server is configured as follows: In response to detecting that the gaze type included in the amblyopia training result is the first gaze type, the training difficulty level is upgraded to obtain the upgraded training difficulty level. In response to detecting that the gaze type included in the amblyopia training result is a second gaze type or a third gaze type, the training difficulty level is downgraded to obtain a downgraded training difficulty level. In response to detecting that the training search time exceeds the preset time threshold, the training difficulty level is downgraded to obtain a downgraded training difficulty level.

5. The system according to claim 1, wherein, The amblyopia training server is configured as follows: In response to determining that the number of training sessions for the amblyopic eye meets a preset ratio condition, the following steps are performed: Based on the training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. Based on the amblyopic eye region and the healthy eye region on the opposite side of the training, the user terminal device is controlled to perform eye-training operations to obtain eye-training results.

6. An amblyopia training method based on object search, applied to the amblyopia training system based on object search as described in any one of claims 1-5, comprising: The training brightness level is determined based on the preset brightness level set and the acquired calibration image set; Generate a set of test images based on user account information and background material set; Based on each test image in the test image set and the training difficulty level, perform the following amblyopia training steps: The test image is subjected to irregular segmentation to obtain the contralateral healthy eye region and the amblyopic eye region; Based on the training brightness level, the contralateral healthy eye region is modified to obtain the training contralateral healthy eye region. Based on the training contralateral healthy eye region and the amblyopic eye region, the user terminal device is controlled to perform amblyopic eye training operations to obtain amblyopic eye training results. Based on the amblyopia training results, the training difficulty level is modified to obtain an updated training difficulty level. The updated training difficulty level is used as the training difficulty level for the next test image, and the amblyopia training steps are continued. In response to the detection that the number of times the amblyopia training step is executed meets a preset number condition, the amblyopia training is terminated, and the training results for each amblyopia are obtained.

7. An electronic device, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors perform the method as described in claim 6.

8. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in claim 6.

Images