Visual search performance evaluation method, device, equipment, medium and program product
By overlaying the target image onto the background image and obtaining the HSV value and image gradient, the visual search performance is quantified using the target model, which solves the problem of inaccurate visual search evaluation in the existing technology and achieves more accurate evaluation results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TSINGHUA UNIVERSITY
- Filing Date
- 2022-09-13
- Publication Date
- 2026-07-21
AI Technical Summary
Existing visual search evaluation methods are not accurate enough and it is difficult to quantify the effectiveness of users in finding target information in complex interfaces.
By acquiring background and target images, superimposing the target image onto different positions of the background image, obtaining the HSV value and image gradient of the candidate image, inputting them into the target model to obtain the average search response time, and using the trained target model to quantify the visual search performance.
It enables the quantification of visual search evaluation results, making the evaluation results closer to the perception of the human visual system and improving the accuracy of the evaluation.
Smart Images
Figure CN115546591B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of human-computer interaction technology, and in particular to a method, apparatus, device, medium and program product for evaluating visual search performance. Background Technology
[0002] With the development of information technology, users often need to find the target information they need in complex interfaces, such as finding the entry point of a link on a large portal website, or finding the icon of a destination on an electronic map interface.
[0003] However, current visual search evaluation methods are not accurate enough. Summary of the Invention
[0004] Therefore, it is necessary to provide a method, apparatus, device, medium, and program product for quantifying visual search evaluation results to address the aforementioned technical problems.
[0005] Firstly, this application provides a method for evaluating visual search performance. The method includes: acquiring a background image and a target image to be evaluated; overlaying the target image onto different positions of the background image to obtain multiple candidate images; acquiring multiple sets of feature data corresponding one-to-one with the multiple candidate images, each set of feature data including image feature data of the corresponding candidate image and image feature data of the target image, wherein the image feature data includes HSV values and image gradients; inputting the multiple sets of feature data into a target model respectively, and obtaining the average search response time based on the output of the target model, the average search response time being used to characterize the performance of searching for the target image from the background image visually.
[0006] In one embodiment, superimposing the target image onto different positions of the background image includes: superimposing the target image onto different positions of the background image using a sliding window method.
[0007] In one embodiment, multiple sets of feature data are input into the target model, and the average search response time is obtained based on the output of the target model, including:
[0008] Each set of feature data is input into the target model to obtain the search response time of the target image at a position in the background image; the average search response time is obtained by calculating the average of the multiple search response times obtained from the multiple sets of feature data.
[0009] In one embodiment, the training process of the target model includes:
[0010] Acquire sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region; input the sample data into the initial detection model for training to obtain the target model.
[0011] In one embodiment, the method further includes:
[0012] The sample background region is extracted from the sample background image according to a pre-set first division rule, which is within the first corresponding viewpoint range; the sample target image is superimposed on the sample background image according to a pre-set second division rule, which is within the second corresponding viewpoint range.
[0013] In one embodiment, acquiring the background image and the target image to be evaluated includes:
[0014] The method involves acquiring multiple background images with different visual display parameters. Correspondingly, the method also includes determining the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0015] Secondly, this application also provides a visual search performance evaluation device. The device includes:
[0016] The first acquisition module is used to acquire the background image and the target image to be evaluated;
[0017] The overlay module is used to overlay the target image onto different positions of the background image to obtain multiple candidate images;
[0018] The second acquisition module is used to acquire multiple sets of feature data corresponding to multiple candidate images. Each set of feature data includes the image feature data of the corresponding candidate image and the image feature data of the target image. The image feature data includes HSV value and image gradient.
[0019] The input module is used to input multiple sets of feature data into the target model. The average search response time is obtained based on the output of the target model. The average search response time is used to characterize the efficiency of searching for the target image from the background image through vision.
[0020] In one embodiment, the overlay module is specifically used to overlay the target image onto different positions of the background image using a sliding window method.
[0021] In one embodiment, the input module is specifically used to input each set of data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image; and to calculate the average of the multiple search response times obtained from the multiple sets of feature data to obtain the average search response time.
[0022] In one embodiment, a training module is also included. The training module is used to acquire sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region. The sample data is input into the initial detection model for training to obtain the target model.
[0023] In one embodiment, the tool further includes a cropping module, which is used to crop a sample background region from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint range; and to overlay a sample target image onto the sample background image according to a pre-set second division rule, wherein the second division rule is within a second corresponding viewpoint range.
[0024] In one embodiment, the first acquisition module is specifically used to acquire multiple background images with different visual display parameters; correspondingly, it also includes a determination module, which is used to determine the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0025] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the aforementioned visual search performance evaluation method.
[0026] Fourthly, this application also provides a computer-readable storage medium. This computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the aforementioned visual search performance evaluation method.
[0027] Fifthly, this application also provides a computer program product. This computer program product includes a computer program that, when executed by a processor, implements the aforementioned visual search performance evaluation method.
[0028] The aforementioned visual search performance evaluation method, apparatus, device, medium, and program products first acquire the background image and target image to be evaluated; then, the target image is superimposed onto different positions of the background image to obtain multiple candidate images; simultaneously, multiple sets of feature data corresponding one-to-one with the multiple candidate images are acquired; finally, the multiple sets of feature data are input into the target model, and the average search response time is obtained based on the output of the target model. The average search response time is used to characterize the performance of visually searching for the target image from the background image. Each set of feature data includes image feature data of the corresponding candidate image and image feature data of the target image. The image feature data includes HSV values and image gradients. By superimposing the target image onto different positions of the background image to obtain multiple candidate images, and then acquiring the HSV values and image gradients of the candidate and target images, and inputting them into the target model, the average search response time for the target image to be searched in the background image is obtained, thus quantifying the visual search evaluation results.
[0029] Furthermore, the evaluation method uses the image's HSV and image gradient, which is closer to the perception of the human visual system. It also considers the influence of the background image on the target image search, resulting in more accurate evaluation results. Attached Figure Description
[0030] Figure 1 This is a flowchart illustrating a visual search performance evaluation method in one embodiment;
[0031] Figure 2 This is a flowchart illustrating a visual search performance evaluation method in one embodiment;
[0032] Figure 3 This is a flowchart illustrating a visual search performance evaluation method in one embodiment;
[0033] Figure 4 This is the definition of the sample background region and the sample target region in another embodiment;
[0034] Figure 5 This is a flowchart illustrating the visual search performance evaluation method in another embodiment;
[0035] Figure 6 This is a flowchart illustrating the visual search performance evaluation method in another embodiment;
[0036] Figure 7 This is a schematic diagram illustrating the optimization process of the visual search performance evaluation method in another embodiment;
[0037] Figure 8 This is a structural block diagram of a visual search performance evaluation device in one embodiment;
[0038] Figure 9This is a structural block diagram of the visual search performance evaluation device in another embodiment;
[0039] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0040] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0041] With the development of information technology, users often need to find the target information they need in complex interfaces, such as finding the entry point of a link on a large portal website or finding the icon of a destination on an electronic map interface. However, current visual search evaluation methods are not accurate enough.
[0042] In view of this, embodiments of this application provide a visual search performance evaluation method. This method obtains multiple candidate images by superimposing the target image onto different positions of the background image, and then obtains the HSV values and image gradients of the candidate images and the target image. These are then input into the target model to obtain the average search response time when the target image is searched in the background image, thereby realizing the quantification of visual search evaluation results.
[0043] In one embodiment, such as Figure 1 As shown, a visual search performance evaluation method is provided. The method is illustrated using a terminal as an example. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. The method includes the following steps:
[0044] Step 101: The terminal acquires the background image and the target image to be evaluated.
[0045] Obtain the background image and target image to be evaluated. The background image does not include the target image. The target image is the image the user wants to search for. For example, the background image could be an electronic map, and the target image could be an icon on the electronic map that the user wants to search for.
[0046] Step 102: The terminal overlays the target image onto different positions of the background image to obtain multiple candidate images.
[0047] A candidate image is obtained when the target image is located at a certain position in the background image. Multiple candidate images are obtained by overlaying the target image onto different positions in the background image. A candidate image is an image of the background image that includes the target image.
[0048] Step 103: The terminal acquires multiple sets of feature data that correspond one-to-one with multiple candidate images.
[0049] Each set of feature data includes image feature data for the corresponding candidate images and image feature data for the target images. The image feature data includes HSV values and image gradients. In image processing, the HSV color space is closer to human color perception; the three color parameters of HSV are hue (H), saturation (S), and lightness (V). Image gradients can serve as an important feature for image recognizability, and the human eye is highly sensitive to image gradients. The HSV values and image gradients of each candidate image are calculated, and the HSV values and image gradients of the target image are also calculated. This data can then be stored in a parameter database.
[0050] Step 104: The terminal inputs multiple sets of feature data into the target model and obtains the average search response time based on the output of the target model.
[0051] The average search response time characterizes the efficiency of visually retrieving the target image from the background image. A longer average search response time indicates lower efficiency, while a shorter average search response time indicates higher efficiency. The target model is a trained model that calculates the search response time for the target image using input image feature data. Multiple sets of feature data are input into the target model, one set at a time, to obtain the search response time when the target image is located at one position in the background image. After inputting multiple sets of feature data, the search response times for the target image at multiple positions in the background image are obtained. The average of these multiple search response times is then calculated to obtain the average search response time for visually retrieving the target image from the background image.
[0052] The aforementioned visual search performance evaluation method first acquires the background image and target image to be evaluated; then, the target image is superimposed onto different positions of the background image to obtain multiple candidate images; simultaneously, multiple sets of feature data corresponding one-to-one with the multiple candidate images are acquired; finally, the multiple sets of feature data are input into the target model, and the average search response time is obtained based on the output of the target model. The average search response time is used to characterize the performance of visually searching for the target image from the background image. Each set of feature data includes the image feature data of the corresponding candidate image and the image feature data of the target image. The image feature data includes HSV values and image gradients. By superimposing the target image onto different positions of the background image to obtain multiple candidate images, and then acquiring the HSV values and image gradients of the candidate and target images, and inputting them into the target model, the average search response time for the target image to be searched in the background image is obtained, thus quantifying the visual search evaluation results.
[0053] Furthermore, the evaluation method uses the image's HSV and image gradient, which is closer to the perception of the human visual system. It also considers the influence of the background image on the target image search, resulting in more accurate evaluation results.
[0054] In the embodiments of this application, in order to obtain the search performance of the target image in the background image, the search response time of the target image at different positions in the background image is obtained by superimposing the target image at different positions in the background image. The specific steps include: superimposing the target image at different positions in the background image using the sliding window method.
[0055] The sliding window method primarily involves sliding a window of a given size through the input background image from left to right and from top to bottom. By sampling the target image at different locations superimposed on the background image, multiple candidate images are obtained. A candidate image is an image of the target image superimposed on a specific location within the background image.
[0056] In the embodiments of this application, such as Figure 2 As shown, multiple sets of feature data are input into the target model, and the average search response time is obtained based on the output of the target model. The specific steps include:
[0057] Step 201: The terminal inputs each set of data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image.
[0058] Each set of feature data includes the HSV values and image gradients of the candidate image and the target image, respectively. The feature data is input into the target model, and the target model outputs the search response time of the target image at the current position in the background image.
[0059] Step 202: The terminal calculates the average of multiple search response times obtained from multiple sets of feature data to obtain the average search response time.
[0060] Multiple sets of feature data are feature data of the target image at multiple positions in the background image. The multiple search response times obtained by calculating multiple sets of feature data are also the search response times of the target image at multiple positions in the background image. The average of the multiple search response times is calculated to obtain the average search response time. The average search response time is the quantitative result of the efficiency of searching for the target image from the background image through vision.
[0061] In the embodiments of this application, the target model needs to be trained using sample data first, so that the target model can obtain the search response time based on the input data, such as... Figure 3 As shown, the specific training process includes:
[0062] Step 301: The terminal acquires sample data.
[0063] The sample data includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region.
[0064] Optionally, the sample background region and the sample target region are determined based on the human eye's ability to process visual information, specifically including the following two cases:
[0065] In the first case, the sample background region is extracted from the sample background image according to the first pre-set first division rule, which is within the first corresponding viewpoint range.
[0066] Among these, the human eye's ability to distinguish fine physical contours and shapes mainly relies on central vision. The macula is the central part of the retina, corresponding to a visual angle of approximately 18 degrees. The first corresponding visual angle is approximately 18 degrees, meaning that the sample background image is divided into 18-degree ranges as the sample background region. Figure 4 Within the large circular area shown.
[0067] In the second case, the sample target image is superimposed on the sample background image according to the pre-set second division rule to obtain the sample target region. The second division rule is within the second corresponding viewpoint range.
[0068] The highest resolution capability of the human eye in processing visual information is based on the fovea, corresponding to a viewing angle of approximately 5 degrees. The second corresponding viewing angle is also approximately 5 degrees, meaning that a 5-degree range within the sample background image is divided into the sample target region, such as... Figure 4 Within the small circular area shown.
[0069] Then, the HSV value and image gradient of the sample background region and the HSV value and image gradient of the sample target region are calculated as input data of the target model. The output data of the target model is the sample search time when the sample target region is in the sample background region. The sample search time is obtained through system experiment. More than 200 participants with different education levels, gender balance and different cultural backgrounds were selected as the test group. (1) The participants were informed of the experimental task, followed the principle of voluntariness, signed informed consent forms, and their information and personal data were registered. (2) Before the formal experiment, each participant practiced 10 times. After completing the training and being able to respond accurately, they entered the formal experiment. (3) In the formal experiment, after every 200 times, the participants were given 20 seconds to rest. After the rest time, they entered the next experimental process. (4) After completing all the experiments, the experiment ended, and the search response time data of each participant was collected.
[0070] Step 302: The terminal inputs the sample data into the initial detection model for training to obtain the target model.
[0071] The above sample data is input into the initial detection model for training to obtain the parameters of the target model, thus obtaining the target model.
[0072] In the above embodiments, by taking into account the human eye's visual angle and collecting real human eye search response time as samples for training the target model, the prediction results of the target model are more accurate.
[0073] Optionally, by acquiring the background image and target image to be evaluated, a quantitative value of the search performance of the target image can be obtained. This quantitative value can then be used to optimize the visual display parameters. Specific steps are as follows: Figure 5 As shown, it includes:
[0074] Step 501: Obtain background images with multiple different visual display parameters.
[0075] Visual display parameters can include parameters such as image contrast, saturation, and brightness. Background images with different visual display parameters are multiple background images with different values for contrast, saturation, or brightness.
[0076] Step 502: Determine the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0077] Based on multiple background and target images with different visual parameters, the evaluation search response time of the target image, i.e., the quantified value of the target image's search performance, is calculated using the methods described above. Based on the target image's search performance, the optimal combination of visual display parameters can be determined. The optimal parameter combination maximizes the target image's search performance, i.e., minimizes the average search time.
[0078] In the above embodiments, by evaluating the search performance of background and target images under different visual parameters, the optimal combination of visual parameters can be found, thereby optimizing the background image to improve the human search response speed. In this way, the user experience can be improved when searching for information on complex interfaces.
[0079] In the embodiments of this application, please refer to Figure 6 The document illustrates a flowchart of a visual search performance evaluation method provided in an embodiment of this application. This visual search performance evaluation method includes the following steps:
[0080] Step 601: The terminal acquires sample data.
[0081] Step 602: The terminal inputs the sample data into the initial detection model for training to obtain the target model.
[0082] Step 603: The terminal acquires the background image and target image to be evaluated.
[0083] Step 604: The terminal overlays the target image onto different positions of the background image to obtain multiple candidate images.
[0084] Step 605: The terminal acquires multiple sets of feature data that correspond one-to-one with multiple candidate images.
[0085] Step 606: The terminal inputs each set of feature data from multiple sets into the target model to obtain the search response time of the target image at a position in the background image.
[0086] Step 607: The terminal calculates the average of multiple search response times obtained from multiple sets of feature data to obtain the average search response time.
[0087] To facilitate readers' understanding of the technical solutions provided in the embodiments of this application, the visual search performance evaluation method and image optimization process of this application are illustrated below with examples. Please refer to... Figure 7 First, a background image and a target image are input. Using a sliding window technique, samples of the target image are taken at different locations on the background image. Then, the HSV values and image gradients of both the background and target images are calculated and input into the target model for computation, yielding the average search response time, which is the search performance of the target image. Next, the optimal visual parameters are obtained by calculating the search performance of the target image under different visual parameters. Finally, the visual parameters of the original image are optimized based on the optimal visual parameters.
[0088] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0089] Based on the same inventive concept, this application also provides a visual search performance evaluation device for implementing the visual search performance evaluation method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the visual search performance evaluation device provided below can be found in the limitations of the visual search performance evaluation method described above, and will not be repeated here.
[0090] In one embodiment, such as Figure 8 As shown, a visual search performance evaluation device 800 is provided, including: a first acquisition module 801, an overlay module 802, a second acquisition module 803, and an input module 804, wherein:
[0091] The first acquisition module 801 is used to acquire the background image and the target image to be evaluated.
[0092] The overlay module 802 is used to overlay the target image onto different positions of the background image to obtain multiple candidate images.
[0093] The second acquisition module 803 is used to acquire multiple sets of feature data corresponding to multiple candidate images. Each set of feature data includes image feature data of the corresponding candidate image and image feature data of the target image. The image feature data includes HSV value and image gradient.
[0094] The input module 804 is used to input multiple sets of feature data into the target model respectively, and obtain the average search response time based on the output of the target model. The average search response time is used to characterize the efficiency of searching for the target image from the background image through vision.
[0095] In the embodiments of this application, the overlay module 802 is specifically used to overlay the target image onto different positions of the background image using a sliding window method.
[0096] In the embodiments of this application, the input module 804 is specifically used to input each set of data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image; and to calculate the average of the multiple search response times obtained from multiple sets of feature data to obtain the average search response time.
[0097] Please refer to Figure 9 This illustrates another visual search performance evaluation device 900 provided in the embodiments of this application. In addition to the modules included in the visual search performance evaluation device 800, the visual search performance evaluation device 900 also includes a training module 901, an interception module 902, and a determination module 903.
[0098] The training module 901 is used to acquire sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region. The sample data is then input into the initial detection model for training to obtain the target model.
[0099] The cropping module 902 is used to crop a sample background region from a sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint; and to overlay a sample target image onto a sample background image according to a pre-set second division rule, wherein the second division rule is within a second corresponding viewpoint.
[0100] In one embodiment of this application, the first acquisition module 801 is specifically used to acquire multiple background images with different visual display parameters. The determination module 903 is used to determine the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0101] Each module in the aforementioned visual search performance evaluation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.
[0102] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 10 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a visual search performance evaluation method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.
[0103] Those skilled in the art will understand that Figure 10The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0104] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0105] The process involves acquiring a background image and a target image to be evaluated; overlaying the target image onto different positions of the background image to obtain multiple candidate images; acquiring multiple sets of feature data corresponding to each candidate image, with each set including image feature data of the corresponding candidate image and image feature data of the target image, where the image feature data includes HSV values and image gradients; inputting the multiple sets of feature data into the target model, and obtaining the average search response time based on the output of the target model, which is used to characterize the effectiveness of searching for the target image from the background image through vision.
[0106] In one embodiment, when the processor executes the computer program, it also performs the following steps: superimposing the target image onto different positions of the background image using a sliding window method.
[0107] In one embodiment, when the processor executes the computer program, it further performs the following steps: inputting each set of feature data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image; calculating the average of the multiple search response times obtained from the multiple sets of feature data to obtain the average search response time.
[0108] In one embodiment, when the processor executes the computer program, it further performs the following steps: acquiring sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region; and inputting the sample data into an initial detection model for training to obtain a target model.
[0109] In one embodiment, when the processor executes the computer program, it further performs the following steps: extracting a sample background region from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint; and superimposing a sample target image onto the sample background image according to a pre-set second division rule to obtain a sample target region, wherein the second division rule is within a second corresponding viewpoint.
[0110] In one embodiment, when the processor executes the computer program, it further performs the following steps: acquiring multiple background images with different visual display parameters; and determining the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0111] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:
[0112] The process involves acquiring a background image and a target image to be evaluated; overlaying the target image onto different positions of the background image to obtain multiple candidate images; acquiring multiple sets of feature data corresponding to each candidate image, with each set including image feature data of the corresponding candidate image and image feature data of the target image, where the image feature data includes HSV values and image gradients; inputting the multiple sets of feature data into the target model, and obtaining the average search response time based on the output of the target model, which is used to characterize the effectiveness of searching for the target image from the background image through vision.
[0113] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: superimposing the target image onto different positions of the background image using a sliding window method.
[0114] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: inputting each set of feature data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image; calculating the average of the multiple search response times obtained from the multiple sets of feature data to obtain the average search response time.
[0115] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region; and inputting the sample data into an initial detection model for training to obtain a target model.
[0116] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: extracting a sample background region from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint; and superimposing a sample target image onto the sample background image according to a pre-set second division rule to obtain a sample target region, wherein the second division rule is within a second corresponding viewpoint.
[0117] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring multiple background images with different visual display parameters; and determining the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0118] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:
[0119] The process involves acquiring a background image and a target image to be evaluated; overlaying the target image onto different positions of the background image to obtain multiple candidate images; acquiring multiple sets of feature data corresponding to each candidate image, with each set including image feature data of the corresponding candidate image and image feature data of the target image, where the image feature data includes HSV values and image gradients; inputting the multiple sets of feature data into the target model, and obtaining the average search response time based on the output of the target model, which is used to characterize the effectiveness of searching for the target image from the background image through vision.
[0120] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: superimposing the target image onto different positions of the background image using a sliding window method.
[0121] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: inputting each set of feature data from multiple sets of feature data into the target model to obtain the search response time of the target image at a position in the background image; calculating the average of the multiple search response times obtained from the multiple sets of feature data to obtain the average search response time.
[0122] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region; and inputting the sample data into an initial detection model for training to obtain a target model.
[0123] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: extracting a sample background region from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint; and superimposing a sample target image onto the sample background image according to a pre-set second division rule to obtain a sample target region, wherein the second division rule is within a second corresponding viewpoint.
[0124] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring multiple background images with different visual display parameters; and determining the optimal combination of visual display parameters based on the average search response time corresponding to each background image.
[0125] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.
[0126] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.
[0127] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0128] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for evaluating the effectiveness of visual search, characterized in that, The method includes: Obtain the background image and target image to be evaluated; The target image is superimposed onto different positions of the background image to obtain multiple candidate images; Multiple sets of feature data corresponding one-to-one with the multiple candidate images are obtained. Each set of feature data includes the image feature data of the corresponding candidate image and the image feature data of the target image. The image feature data includes HSV value and image gradient. The multiple sets of feature data are input into the target model respectively, and the average search response time is obtained based on the output of the target model. The average search response time is used to characterize the efficiency of searching for the target image from the background image through vision. The acquisition of the background image and target image to be evaluated includes: Acquire background images with multiple different visual display parameters; Correspondingly, the method also includes: The optimal combination of visual display parameters is determined based on the average search response time corresponding to each of the background images; The target model is trained based on sample data, which includes a sample background region and a sample target region. The method further includes: The sample background region is extracted from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewing angle range; the first corresponding viewing angle range is determined based on the central visual acuity of the human eye. The sample target image is superimposed on the sample background image according to a pre-set second division rule to obtain the sample target region. The second division rule is within a second corresponding viewpoint range, and the second corresponding viewpoint range is determined according to the highest resolution of the human eye.
2. The method according to claim 1, characterized in that, The step of overlaying the target image onto different positions of the background image includes: The target image is superimposed on different positions of the background image using a sliding window method.
3. The method according to claim 1, characterized in that, The step of inputting the multiple sets of feature data into the target model respectively, and obtaining the average search response time based on the output of the target model, includes: Each set of data from the multiple sets of feature data is input into the target model to obtain the search response time of the target image at a position in the background image; The average search response time is obtained by calculating the average of the multiple search response times obtained from the multiple sets of feature data.
4. The method according to claim 1, characterized in that, The training process of the target model includes: Acquire sample data, which includes the HSV value and image gradient of the sample background region, the HSV value and image gradient of the sample target region, and the sample search response time corresponding to the sample background region and the sample target region; The sample data is input into the initial detection model for training to obtain the target model.
5. The method according to claim 1, characterized in that, The visual display parameters include the image's contrast, saturation, and brightness.
6. The method according to claim 1, characterized in that, The first corresponding angle is 18 degrees, and the second corresponding angle is 5 degrees.
7. A visual search performance evaluation device, characterized in that, The device includes: The first acquisition module is used to acquire the background image and the target image to be evaluated; An overlay module is used to overlay the target image onto different positions of the background image to obtain multiple candidate images; The second acquisition module is used to acquire multiple sets of feature data corresponding one-to-one with the multiple candidate images. Each set of feature data includes image feature data of the corresponding candidate image and image feature data of the target image. The image feature data includes HSV value and image gradient. The input module is used to input the multiple sets of feature data into the target model respectively, and to obtain the average search response time based on the output of the target model. The average search response time is used to characterize the efficiency of searching for the target image from the background image through vision. The first acquisition module is specifically used to acquire multiple background images with different visual display parameters; The determination module is used to determine the optimal combination of visual display parameters based on the average search response time corresponding to each of the background images; The target model is trained based on sample data, which includes a sample background region and a sample target region. The cropping module is used to crop the sample background region from the sample background image according to a pre-set first division rule, wherein the first division rule is within a first corresponding viewpoint range, and the first corresponding viewpoint range is determined based on the central visual acuity of the human eye. The sample target image is superimposed on the sample background image according to a pre-set second division rule to obtain the sample target region, wherein the second division rule is within a second corresponding viewpoint range, and the second corresponding viewpoint range is determined based on the highest resolution of the human eye.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.