AI automatic game software testing method and system

Abstract

The invention discloses an AI automatic game software testing method and system, and relates to the field of game software automatic testing, the AI automatic testing system generates a test task according to a test case configuration table, a test assistant executes an instruction of the test task, and obtains a current screen image of game software; according to the method, images are identified through various CV identification models, various CV information is obtained, the game interface state is judged after fusion analysis is conducted on the various CV information, corresponding operation steps are executed on game software according to a test task to simulate various environment variables, and the game stability is evaluated by monitoring the change of a game interface through the CV models. After the test task is finished, the test console automatically generates a test report; according to the automatic test system, the terminal game software image can be recognized through the CV model to obtain various CV information, the test task is automatically completed in cooperation with the test case configuration table, the test effect of the game software is improved, and the test cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of game software testing, and in particular to an AI automated game software testing method and system. Background Art

[0002] With the popularity of smart phones and the improvement of their performance, the mobile game market has grown rapidly and has become an important part of the global mobile application market. Players have high requirements for the game's graphics quality, fluency, operational response and stability, which puts higher standards on the performance and quality of the game. Game software testing is a very important and indispensable part of the game development project. The effectiveness of the testing work is directly related to the quality of the game software product.

[0003] Existing game software testing technologies are generally divided into white box testing and black box testing. White box testing, also known as transparent box testing, is code-based testing that allows testers to deeply access the internal details of the software program and conduct targeted testing on the program code and logic modules. Black box testing does not go deep into the interior of the software program. It treats the test object as a black box and tests the functions of the game software from the player's perspective.

[0004] The prior art has the following deficiencies:

[0005] Existing white-box testing methods all use invasive technical methods to perform testing, that is, it is necessary to implant relevant test code in the program code of the game software, or the game software needs to provide an open interface to perform testing. Although the software program code can be tested accurately, the game cannot be tested from the perspective of a real player. In actual engineering development, it is often the case that a function passes the white-box test, but there are problems when the real player plays on the terminal. Although black-box testing is performed from the perspective of a real player, it currently usually requires professional testers to perform testing. Manual testing usually relies on human autonomy. When too much repetitive test content is performed, visual fatigue is likely to occur, resulting in missed tests and wrong tests, which ultimately leads to the release of a version with BUGs online, and labor costs rise linearly with the iteration of game functions. Summary of the invention

[0006] The purpose of the present invention is to provide an AI automated game software testing method and system, which generates test tasks through a test case configuration table, and then uses a variety of computer vision (CV) models with different functions to identify the terminal screen image to obtain a variety of CV information and fuse them, automatically complete the black box test content, and perform interactive testing from the player's perspective and environment. There is no need to invade the game software program code and manual secondary testing, thereby improving the test effect and reducing the test cost.

[0007] In order to achieve the above object, the present invention provides the following technical solution: an AI automated game software testing method, the testing method comprising the following steps:

[0008] The automated test console parses the test case through the test case configuration table, extracts the content that needs to be tested in the game software, generates the corresponding initial test task according to the test case configuration table, and configures the CV model. The automated test console sends a startup instruction to the test assistant APP installed on the terminal. The test assistant initializes the test environment according to the initial test task and starts to execute the CV information request of the automated test console;

[0009] The test assistant captures the current screen image of the game software. The CV model recognition server recognizes the screen image through multiple CV models to obtain multiple CV information. After fusing and analyzing the multiple CV information, the game interface status, player operation and environmental feedback are determined.

[0010] The test assistant executes the corresponding operation steps on the game software according to the test case configuration table, simulates multiple environmental variables during the operation, monitors the changes of the game interface in various simulated environments through the CV model, and evaluates the stability of the game. After the test is completed, the automated test console automatically generates a test report.

[0011] In a preferred embodiment, multiple environmental variables are simulated during operation to verify the response of the game from the player's perspective, and the game stability is evaluated after monitoring the changes in the game interface under various simulated environments through the CV model, including the following steps:

[0012] Simulate multiple environmental variables during operation, including network latency, screen brightness, battery level, terminal device temperature, storage space, and volume settings;

[0013] Obtain network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor;

[0014] Substitute the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor into the linear analysis algorithm to calculate the overall abnormality coefficient of the game under different environmental variables;

[0015] The overall abnormality coefficient is compared with the preset coefficient threshold. The coefficient threshold is used to evaluate whether the game stability is good or poor. If the overall abnormality coefficient is greater than the coefficient threshold, the game stability is evaluated to be poor. If the overall abnormality coefficient is less than or equal to the coefficient threshold, the game stability is evaluated to be excellent.

[0016] In a preferred embodiment, the network delay operation abnormality factor, the screen brightness operation abnormality factor, the power level operation abnormality factor, the terminal device temperature operation abnormality factor, the storage space operation abnormality factor, and the volume setting operation abnormality factor are substituted into the linear analysis algorithm to calculate the overall abnormality coefficient of the game under different environmental variables. The expression is: Where YZx is the overall abnormality coefficient, {y1, y2, y3, y4, y5, y6} are the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, volume setting operation abnormality factor, {w1, w2, w3, w4, w5, w6} are the regression coefficients of the abnormal factors of each environmental variable operation, and the regression coefficients are greater than 0.

[0017] In a preferred embodiment, the automated test console parses the test case through the test case configuration table, extracts the content that needs to be tested in the game software, generates the corresponding initial test task according to the test case configuration table, and configures the CV model, including the following steps:

[0018] The automated test console reads information from the predefined test case configuration table, which records the content of each test case, including the test process, test objectives, and operation steps;

[0019] Parse the contents of the test case configuration table and extract each test case in the test case configuration table one by one. Each test case contains the steps of the test process, target expected results, test conditions and execution standards;

[0020] Based on the definition of the test case, the automated test console identifies the test process and test objectives of the test case. The process includes operation steps, and the objective is the expected result of the test.

[0021] After parsing and extracting the process and target, the automated test console generates the corresponding initial test task, which contains the description of each test step, including the execution order, expected results, and operation type information;

[0022] According to the process and objectives of each test case, the CV (computer vision) information that needs to be verified is identified, and different CV models are used to detect different CV information.

[0023] In a preferred embodiment, the automated test console sends a start instruction to the test assistant, the test assistant initializes the test environment according to the initial test task, and executes the CV information request of the automated test console, including the following steps:

[0024] The automated test console sends a start command to the test assistant, instructing it to start the test process and prepare for the test. The command includes the test case ID, task information, and CV model requirements required to start the test.

[0025] After receiving the start command, the test assistant first checks its own status. If any abnormality is found, the test assistant returns a status report and notifies the automated test console to make corrections or restart the command;

[0026] The test assistant loads the corresponding task configuration according to the initial test task information contained in the startup instruction, and sets up the corresponding test environment according to the requirements of the initial test task. The environment initialization involves simulating network conditions, setting screen resolution, adjusting brightness, and configuring sound.

[0027] After the environment is prepared, the test assistant executes the CV information request of the automated test console, which contains the identification requirements.

[0028] In a preferred embodiment, the test assistant captures the current screen image of the game software, and the CV model recognition server recognizes the screen image through multiple CV models to obtain multiple CV information, including the following steps:

[0029] The test assistant APP runs on the terminal device and captures the screen image of the current game interface according to the request of the automated test console. The image capture includes the entire screen content, so that the automated test system can obtain UI elements, text and scene background information;

[0030] The test assistant packages the captured screen image data and transmits it to the CV model recognition server through the network. If packet loss or errors occur during the image data transmission process, the test assistant retransmits the image until the server successfully receives it.

[0031] After receiving the image data, the CV model recognition server loads the CV model according to the test requirements. The CV model includes the target detection model, the text recognition model, the scene classification model, and the instance segmentation model.

[0032] The object detection model is used to detect UI elements on the screen and identify their position and size; the text recognition model is used to identify text information in the game interface; the scene classification model is used to analyze the background and overall layout of the game scene and identify the characteristics of different scenes; the instance segmentation model is used to accurately identify the image area that needs to be analyzed at the pixel level;

[0033] Each CV model is called in turn to analyze the screen image, and each CV model extracts information at different levels of the image;

[0034] The text recognition model extracts the text content on the screen and converts it into a readable data format. The scene classification model identifies the overall layout and background features of the image to determine whether it meets the specified game scene requirements. The instance segmentation model can more accurately locate the image area that needs to be recognized.

[0035] The test assistant performs fusion analysis on the information obtained from each CV model, packages the CV information obtained from the analysis, and returns it to the automated test console.

[0036] In a preferred embodiment, the test assistant performs corresponding operation steps on the game software according to the test case configuration table, including the following steps:

[0037] The test assistant receives and parses the test case configuration table from the automated test console, and determines the operation steps to be performed according to the task instructions in the test case configuration table;

[0038] The test assistant uses the recognition results of the target detection model to confirm the position and status of the UI elements that need to be interacted with in the interface, ensuring that the target elements are visible and in an interactive state. Otherwise, an exception is recorded and the operation is skipped.

[0039] According to the operation type specified in the test case configuration table, the test assistant performs corresponding user interaction behaviors on the terminal device. User interaction behaviors include clicking, sliding, and long pressing to simulate the user's operation habits.

[0040] AI automated game software testing system, including initialization configuration module, image recognition module, testing module, and report generation module;

[0041] Initialization configuration module: parse the test case through the test case configuration table, extract the content that needs to be tested in the game software, generate the corresponding initial test task according to the test case configuration table, and configure the CV model;

[0042] Image recognition module: Initializes the test environment according to the initial test task, and sends a CV information request to the test assistant APP on the terminal device. The test assistant captures the current screen image of the game software, recognizes the screen image through multiple CV models, obtains multiple CV information, and determines the game interface status, player operation and environmental feedback after fusion analysis of multiple CV information;

[0043] Testing module: Execute corresponding operation steps on the game software according to the test case configuration table, simulate multiple environmental variables during the operation, and evaluate the game stability after monitoring the changes of the game interface in various simulated environments through the CV model;

[0044] Report generation module: After the test is completed, the automated test console automatically generates a test report, which includes game evaluation results, error logs, CV information, operation steps and environment variables.

[0045] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0046] The present invention generates corresponding initial test tasks according to the test case configuration table, and configures the CV model. The test assistant executes the current screen image of the game software, and the CV model recognition server recognizes the screen image through multiple CV models to obtain multiple CV information. The multiple CV information is integrated and analyzed to determine the state of the game interface. The test assistant performs corresponding operation steps on the game software according to the test case configuration table, simulates multiple environmental variables during the operation, and evaluates the game stability after monitoring the changes of the game interface under various simulated environments through the CV model. After the test is completed, the automated test console automatically generates a test report. The test system recognizes the terminal screen image through multiple different CV models to obtain multiple CV information and integrates them, and then automatically completes the test content through the test case configuration table, and tests from the player's perspective and environment, without invading the game software code and manual secondary testing, thereby improving the test effect and reducing the test cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0048] Figure 1 is a system architecture diagram of the present invention, Figure 2 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Example 1: Please refer to Figure 1 and Figure 2 As shown, the AI ​​automated game software testing method and system of this embodiment, the testing method includes the following steps:

[0051] The automated test console parses the test case through the test case configuration table, extracts the content that the game software needs to test, including the process and the target, generates the corresponding initial test task according to the test case configuration table, and configures the CV (computer vision) model. The automated test console sends a startup instruction to the test assistant to ensure that the test assistant is active to support the next test process. The test assistant initializes the test environment according to the initial test task and executes the CV information request of the automated test console. The test assistant executes the capture of the current screen image of the game software. The CV model server uses a variety of CV models (for example, target detection, text recognition, scene classification, instance analysis, etc.) Cutting, etc.) to identify the screen image and obtain various CV information. After fusing and analyzing the various CV information, the game interface status, player operation and environmental feedback are determined. The test assistant executes corresponding operation steps on the game software according to the test case configuration table. The operation steps include different user interaction behaviors such as clicking, sliding, and long pressing. During the operation, multiple environmental variables are simulated to verify the response of the game from the player's perspective. The game stability is evaluated after monitoring the changes of the game interface in various simulated environments through the CV model. After the test, the automated test console automatically generates a test report. The test report includes game evaluation results, error logs, CV information, operation steps and environmental variables.

[0052] This application generates the corresponding initial test tasks according to the test case configuration table, and configures the CV model. The test assistant executes to capture the current screen image of the game software. The CV model recognition server recognizes the screen image through multiple CV models to obtain multiple CV information. After fusing and analyzing the multiple CV information, the game interface status is determined. The test assistant performs corresponding operation steps on the game software according to the test case configuration table, simulates multiple environmental variables during the operation, and evaluates the game stability after monitoring the changes in the game interface in various simulated environments through the CV model. After the test is completed, the automated test console automatically generates a test report. The test system recognizes the terminal screen image through a variety of different CV models to obtain multiple CV information and fuse them, and then automatically completes the test content through the test case configuration table, and tests from the player's perspective and environment, without invading the game software code and manual secondary testing, improving the test effect and reducing the test cost.

[0053] Embodiment 2: The automated test console parses the test case through the test case configuration table, extracts the content that needs to be tested for the game software, including the process and the target, generates the corresponding initial test task according to the test case configuration table, and configures the CV (computer vision) model, including the following steps:

[0054] The automated test console reads information from a predefined test case configuration table. The configuration table is usually a structured data file (such as JSON, XML, or Excel), which records the content of each test case in detail, including the test process, test objectives, and operation steps.

[0055] The console parses the contents of the configuration table and extracts each test case in the configuration table one by one. Each test case contains the steps of the test process, the target expected results, the test conditions and the execution standards. The console saves this information in a structured manner for subsequent use.

[0056] Based on the definition of the test case, the automated test console identifies the test process and test objectives of the test case. The process includes specific operation steps (for example, opening the game, clicking a button, switching scenes, etc.), and the goal is the expected result of the test (such as checking login success, confirming button response, verifying scene switching, etc.).

[0057] The console further analyzes the test requirements of each test case to determine the specific content that needs to be monitored and verified. The test requirements may involve the existence of interface elements, the correctness of text information, the smoothness of animations, etc. This step helps the console understand the core checkpoints of the test case and ensure that the testing process focuses on the correct game function modules and interface elements.

[0058] After parsing and extracting the process and goals, the automated test console generates the corresponding initial test tasks. The initial test tasks contain detailed instructions for each test step, including execution order, expected results, operation type (click, slide, input, etc.), etc. These tasks will be automatically executed in the specified order during subsequent testing.

[0059] Based on the flow and goals of each test case, the console identifies the CV (computer vision) model that needs to be configured. Different models are used to detect different game elements. For example, button recognition may require an object detection model, text verification requires a text recognition model, and scene recognition requires a scene classification model. The console records these requirements so that the appropriate CV model can be loaded in subsequent steps.

[0060] The automated test console sends a start command to the test assistant to ensure that the test assistant is active to support the subsequent test process. The test assistant initializes the test environment according to the initial test task and executes the CV information request of the automated test console, including the following steps:

[0061] The automated test console sends a start command to the test assistant, instructing it to start the test process and prepare for the test. The command contains the basic information required to start the test, such as the test case ID, task information, CV model requirements, etc.

[0062] After receiving the start command, the test assistant first checks its own status to ensure that all system services (such as network connection, data processing module, etc.) are in normal status. If any abnormality is found, the test assistant will return a status report to notify the automated test console to make corrections or restart the command.

[0063] The test assistant loads the corresponding task configuration according to the initial test task information contained in the startup instruction. The initial test task includes specific operation steps, expected test results, interface elements to be monitored, etc. The server saves these tasks in memory for subsequent execution.

[0064] The test assistant sets up the corresponding test environment according to the requirements of the initial test task. Environment initialization may involve simulating network conditions, setting screen resolution, adjusting brightness, configuring sound, etc. to simulate the usage conditions of real users. Ensure that all necessary environment variables are set to provide stable and controllable conditions for testing.

[0065] After the environment is prepared, the automated test console sends a CV information request to the client terminal device (with the test assistant APP installed). The CV information request contains specific recognition requirements, such as the interface elements, text content, target scenes, etc. that need to be recognized. This information guides the test assistant APP to perform corresponding image capture and processing.

[0066] The automated test console ensures that the test assistant is active and completes the initialization of the test environment, providing support for the subsequent test process. The test assistant app then captures the screen image based on the CV information request to provide the necessary data for CV model analysis.

[0067] The test assistant captures the current screen image of the game software, and the CV model recognition server recognizes the screen image through multiple CV models (for example, target detection, text recognition, scene classification, instance segmentation, etc.) to obtain multiple CV information, including the following steps:

[0068] The test assistant app runs on the terminal device and captures the screen image of the current game interface based on the request of the automated test console. The image capture includes the entire screen content to ensure that the automated test system can obtain all relevant UI elements, text, scene background and other information. The test assistant can capture screen images in a timed or triggered manner during specific game operations or scene switching according to the requirements of different test cases.

[0069] The test assistant packages the captured screen image data and transmits it to the CV model recognition server through the network. To ensure data integrity, a data verification mechanism may be applied. If packet loss or errors occur during image data transmission, the test assistant will retransmit the image until the server successfully receives it.

[0070] After receiving the image data, the CV model recognition server loads the appropriate CV (computer vision) model according to the test requirements. The functions of different models are as follows:

[0071] Object detection model: used to detect UI elements (such as buttons and icons) on the screen and identify their position and size.

[0072] Character recognition (OCR) model: used to recognize text information in the game interface (such as prompt text, scores, task information, etc.) to determine whether the text content is displayed correctly.

[0073] Scenario classification model: used to analyze the background and overall layout of the game scene and identify the characteristics of different scenes to confirm whether they meet the requirements of the test case.

[0074] Instance segmentation model: used to segment the irregular target area that needs to be identified in the screen image to perform accurate target detection.

[0075] The CV model recognition server calls each CV model in turn to analyze the screen image, and each model extracts information at different levels of the image. For example, the object detection model will identify the location of interface elements (such as button location, sliding area, etc.) and record the element status (such as whether it is pressed).

[0076] The text recognition model extracts the text content on the screen and converts it into a readable data format for subsequent comparison and verification. The scene classification model identifies the overall layout and background features of the image to determine whether it meets the specified game scene requirements.

[0077] The automated test system integrates and analyzes the information obtained from each CV model. For example, by combining the results of object detection and text recognition, it can be confirmed whether the state of a button and its corresponding prompt text are correct. Through the scene classification results, it is further verified whether the current scene conforms to the game process specified by the test case (such as scene switching, level loading, etc.).

[0078] The automated testing system can obtain rich CV information through joint analysis of multiple CV models, accurately identify the state of the game interface, and provide reliable judgment basis and data support for the testing process.

[0079] After integrating and analyzing various CV information, the game interface status, player operation, and environmental feedback are determined, including the following steps:

[0080] The CV model recognition server aggregates the information extracted from multiple CV models such as target detection, text recognition, and scene classification. This information includes the location and status of UI elements, text content, scene background features, etc., providing a data basis for subsequent fusion analysis.

[0081] Standardize the format of data from different CV models for unified processing. For example, convert the coordinate information of target detection and the text content of character recognition into a unified data structure. Check and clean up noise or inaccurate results in the data. For example, if a test result does not meet expectations, it can be corrected by comparing other information.

[0082] Match the interface elements identified by the object detection model with the results of text recognition and scene classification. For example: confirm whether the position and state of a button or icon are consistent with the displayed text description. Ensure that the UI elements and background features in a specific scene match (for example, the background changes when switching levels). Through this matching, you can check whether the layout of elements in the game interface meets the test requirements.

[0083] Based on the fused CV information, the automated testing system makes a comprehensive judgment on the state of the game interface. For example, it checks whether the key elements of the game interface are in the correct display state. It confirms whether the interface changes are consistent with the current operation steps, for example, whether the game scene is correctly entered after clicking the "Start" button. Through this state judgment, the automated testing system can detect whether the game interface responds to user operations as expected.

[0084] All fusion analysis results are organized into structured data and returned to the automated test console. The results include the current interface status, player operation response, environmental feedback, and abnormal information. These analysis results will be used to determine the process of automated testing and decide whether to continue the next operation or generate a final test report.

[0085] The automated testing system can comprehensively judge the state of the game interface, player operations, and environmental feedback, providing accurate support for the testing process. At the same time, fusion analysis helps identify potential problems and provides an effective basis for improving the quality of game software.

[0086] The test assistant performs corresponding operation steps on the game software according to the test case configuration table. The operation steps include different user interaction behaviors such as clicking, sliding, and long pressing, including the following steps:

[0087] The test assistant receives and parses the test case configuration table from the automated test console, and determines the specific operation steps to be performed based on the task instructions in the configuration table. The configuration table includes detailed information such as the type of each operation (such as click, slide, long press), the location of the target element, the timing of the operation, and the expected result.

[0088] The test assistant uses the recognition results of the target detection model to confirm the position and status of the UI elements that need to be interacted with in the interface. Ensure that the target elements are visible and in an interactive state (such as the button is not blocked, the sliding area is available, etc.), otherwise record the exception and skip the operation.

[0089] According to the operation type specified in the configuration table, the test assistant performs the corresponding user interaction behavior on the terminal device, including: Click: Touch the screen at the specified location to implement a single-click operation, which is used to trigger buttons, select options, etc. Slide: Slide on the screen along the specified path (starting point and end point) to scroll pages, move objects, etc. Long press: Keep touching the target location for a certain period of time to trigger the long press menu or special functions. The test assistant accurately controls the strength and duration of the operation to simulate the operating habits of real users.

[0090] After each operation is performed, the test assistant captures the screen image in real time and sends it to the CV model recognition server to obtain operation feedback. The automated test system analyzes the feedback image through the CV model to determine whether the operation has produced the expected effect (such as whether the interface switches, whether the button responds, etc.).

[0091] During the operation, multiple environmental variables are simulated to verify the game response from the player's perspective. The game stability is evaluated after monitoring the changes in the game interface under various simulated environments through the CV model, including the following steps:

[0092] Simulate multiple environmental variables during the operation, including network delay, screen brightness, power level, terminal device temperature, storage space, volume setting, etc., and obtain network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, volume setting operation abnormality factor;

[0093] Substitute the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor into the linear analysis algorithm to calculate the overall abnormality coefficient of the game under different environmental variables. The expression is: Where, YZx is the overall abnormality coefficient, {y1, y2, y3, y4, y5, y6} are the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, volume setting operation abnormality factor, {w1, w2, w3, w4, w5, w6} are the regression coefficients of the abnormal factors of each environmental variable operation, and the regression coefficients are greater than 0;

[0094] The larger the overall abnormality coefficient, the worse the stability of the game. The overall abnormality coefficient is compared with the preset coefficient threshold. The coefficient threshold is used to evaluate whether the game stability is good or bad. If the overall abnormality coefficient is greater than the coefficient threshold, the game stability is evaluated as poor. If the overall abnormality coefficient is less than or equal to the coefficient threshold, the game stability is evaluated as good.

[0095] When the present invention is used, the logical factors of the overall abnormality coefficient are composed of: taking the influence of abnormal factors of different environmental variables as an example, the first is the index, that is, the factor that causes the stability of the game to decline (the present invention refers to the influence of abnormal factors of different environmental variables on the stability of the game); the second is the weight of these indicators, that is, the proportion of each major influencing data when it is generated; the third is the operation equation, that is, what kind of mathematical operation process is used to obtain the result, and the overall abnormality coefficient is obtained by calculating the indicators with their respective weights through the operation equation.

[0096] The main influencing data obtained from the sample were transformed and processed into a data language recognized by computer software; secondly, these evaluation factors were analyzed by Logistic regression using SPSS software to screen out factors and their weights that are significantly correlated with the results; thirdly, the evaluation factors and weights were brought into the Logistic regression equation for calculation to obtain the results, which are as follows:

[0097] First, ensure the integrity of the main effect data, handle missing values ​​and outliers, and convert the data into a format that SPSS software can recognize. Usually, the data is stored in .csv, .xlsx and other formats, and then imported into SPSS. Open SPSS software, import the processed data file, and transform the variables as needed. For example, for continuous variables, standardize or normalize them. Select the "Analyze" menu, then select the "Binary Logistic" option under "Regression". In the dialog box, add the dependent variable (outcome) and the independent variable (main effect data) to the corresponding boxes. SPSS will fit the Logistic regression based on the selected variables. Return to the model. In the output results, you will see the coefficients, standard errors, p-values ​​and other information of the model. Check the coefficients and p-values ​​in the output results to determine which variables are significantly correlated with the results. Usually, a p-value less than 0.05 is considered significant. While fitting the model, use variable selection methods, such as stepwise regression, to help screen the most relevant factors. According to the coefficients of the Logistic regression model, the size of the coefficient reflects the degree of influence of each factor on the result, and the positive and negative signs of the coefficients indicate the direction of the influence. After obtaining the significant factors and their coefficients, the Logistic regression equation is obtained, which is used to calculate the probability of each sample and then predict the results.

[0098] The logic for obtaining the network delay operation abnormality factor is:

[0099] Operate the game in a high-latency network environment, obtain the number of game freezes, and divide the number of freezes in the high-latency network environment by the monitoring time to obtain the network delay operation abnormality factor. The larger the network delay operation abnormality factor, the worse the game stability.

[0100] The logic for obtaining the screen brightness operation abnormality factor is:

[0101] Get the contrast decrease rate of the game interface under different screen brightnesses, subtract the current visibility of the game interface at the corresponding screen brightness from the standard contrast of the game interface at the corresponding screen brightness to get the visibility difference, divide the visibility difference by the standard contrast of the game interface at the corresponding screen brightness to get the contrast decrease rate, sum the contrast decrease rates of all screen brightnesses to get the screen brightness operation abnormality factor. The larger the screen brightness operation abnormality factor, the worse the game stability.

[0102] The logic for obtaining the power level operation abnormality factor is:

[0103] Get the frame rate drop rate of the game interface under different power levels, subtract the game interface frame rate of the current power level from the game standard frame rate to get the frame rate difference, divide the frame rate difference by the game standard frame rate to get the frame rate drop rate, sum the frame rate drop rates under all power levels to get the power level operation abnormality factor. The larger the power level operation abnormality factor, the worse the game stability.

[0104] The logic for obtaining the abnormal temperature operation factor of the terminal device is:

[0105] Get the number of freezes in the game interface at different temperatures, and compare the number of freezes under all temperature conditions with the monitoring time to get the abnormal temperature operation factor of the terminal device. The larger the abnormal temperature operation factor of the terminal device, the worse the game stability.

[0106] The logic for obtaining the storage space operation abnormality factor is as follows:

[0107] Get the number of times the game interface fails to load when the storage space is insufficient, and divide the number of times the game interface fails to load by the monitoring time to get the storage space operation abnormality factor. The larger the storage space operation abnormality factor is, the worse the game stability is.

[0108] The logic for obtaining the abnormal factor of the volume setting operation is:

[0109] Get the number of times the game's sound effects are missing under different volume conditions, and divide the number of times the game's sound effects are missing under all volume conditions by the monitoring duration to get the volume setting operation abnormality factor. The larger the volume setting operation abnormality factor, the worse the game stability.

[0110] After the test is completed, the automated test console automatically generates a test report, which includes game evaluation results, error logs, CV information, operation steps, and environment variables, etc. The test report includes the following steps:

[0111] Summarize all data recorded during the test, including the execution status of each test step, operation results, abnormal conditions, etc. Collect the settings of environmental variables (such as network latency, power level, screen brightness, etc.) and their impact on the game interface and operation response. Organize the analysis results of the CV model, including the identified interface status, player operation feedback, and changes under environmental simulation.

[0112] Evaluate the overall stability and response speed of the game based on test data, and its performance under different environmental variables. Calculate the performance indicators of the game in different test scenarios (such as average response time, frame rate change, operation delay, etc.) and give a quantitative evaluation. Combined with CV information, generate an evaluation of the game's interface stability and identify common display problems or abnormal behaviors (such as freezes, image distortion, etc.).

[0113] Record all errors that occur during testing, including crashes, freezes, UI display anomalies, unresponsive functions, etc. Mark the environmental conditions (such as specific network status, brightness, power level, etc.) and reproduction steps for each error so that the development team can quickly locate the problem. Include detailed error logs and screenshots to assist in analyzing the root cause of the problem.

[0114] Organize the recognition results of the CV model and aggregate different CV information (such as target detection, text recognition, scene classification, etc.) into a unified description of the game interface status. Analyze the performance changes of interface elements under different operation steps and environmental variables, and judge the stability and consistency of the interface in various environments. If the CV model detects an interface anomaly, mark the corresponding steps and environmental conditions for further inspection.

[0115] According to the test case configuration table, record the execution details of each operation step, including the operation type (click, slide, long press, etc.), operation object, and execution result. Record the execution effect of each step, such as whether there is a significant operation delay, or whether the operation causes interface changes. Associate each operation step with the corresponding environment variables and interface status so that there is a clear context when reviewing the test process.

[0116] Record the environment variable settings used in the test and their changes in detail (such as switching from high brightness to low brightness, or from no network latency to high latency). Mark the corresponding environment variables for each test scenario so that the evaluation results can be compared for different environmental conditions. Clearly indicate the impact of environmental variables on game performance and interface stability in the report, and analyze the adaptability under different conditions.

[0117] Convert test data into visual charts, such as response time curves, frame rate fluctuation charts, error distribution charts, etc., to quickly understand the performance of the game. Use charts to show the impact of different environmental variables on the game, and help find under which conditions the game shows large fluctuations. Generate statistical analysis for key performance indicators, such as average response time, crash rate, etc., to provide data support for optimization decisions.

[0118] Summarize the strengths and weaknesses of the game based on the test data, and point out the main problems that occur in specific environments. Propose improvement suggestions, such as optimizing performance under low battery or high network latency conditions, improving the adaptability of certain interface elements, etc. Provide key areas for subsequent improvements for the development and testing teams.

[0119] Integrate all test data, analysis results, error logs, CV information, charts, and improvement suggestions into the final test report document. Present the report content in a structured way so that readers can quickly find key issues and related test data. Export the report to PDF or other formats for easy sharing and archiving. Through these steps, the report generated by the automated test console comprehensively covers the game's evaluation results and problem records, provides clear data support and improvement directions, and helps the development team improve the quality of the game and user experience.

[0120] Embodiment 3: The AI ​​automated game software testing system of this embodiment includes an initialization configuration module, an image recognition module, a testing module, and a report generation module;

[0121] Initialization configuration module: parse the test case through the test case configuration table, extract the content that needs to be tested in the game software, including the process and the goal, generate the corresponding initial test task according to the test case configuration table, and configure the CV (computer vision) model. The initial test task and the CV (computer vision) model are sent to the image recognition module, and the test case configuration table is sent to the test module;

[0122] Image recognition module: Initializes the test environment according to the initial test task, and sends a CV information request to the test assistant APP on the terminal device. The test assistant captures the current screen image of the game software, recognizes the screen image through multiple CV models (for example, target detection, text recognition, scene classification, instance segmentation, etc.), obtains multiple CV information, and determines the game interface status, player operation and environment feedback after fusion analysis of multiple CV information. The CV information is sent to the report generation module;

[0123] Testing module: Execute corresponding operation steps on the game software according to the test case configuration table. The operation steps include different user interaction behaviors such as clicking, sliding, and long pressing. Simulate multiple environmental variables during the operation process to verify the response of the game from the player's perspective. Use the CV model to monitor the changes in the game interface under various simulated environments and evaluate the game stability. The operation steps, environmental variables, and game evaluation results are sent to the report generation module;

[0124] Report generation module: After the test is completed, the automated test console automatically generates a test report, which includes game evaluation results, error logs, CV information, operation steps and environment variables.

[0125] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0126] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0127] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0128] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0129] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. AI automated game software testing method and system, characterized by: The test method comprises the following steps: Test case configuration table parsing step, extracting the content that needs to be tested in the game software, generating the corresponding initial test task according to the test case configuration table, and configuring the CV model. The automated test console sends a startup instruction to the test assistant APP on the terminal device. The test assistant initializes the test environment according to the initial test task and starts to execute the task instruction; The test assistant performs the game software screen recognition step. The test assistant captures the current screen image of the game software and sends it to the CV server. The CV server recognizes the screen image through multiple CV models and obtains multiple CV information. After fusing and analyzing the multiple CV information, the game interface status, player operation and environmental feedback are determined. The test assistant executes the test task instruction steps. The test assistant performs the corresponding operation steps according to the test task content generated by the test case configuration table. During the operation, it simulates multiple environmental variables, monitors the changes in the game interface under various simulated environments through the CV model, and evaluates the game stability. After the test is completed, the automated test console automatically generates a test report.

2. The AI ​​automated game software testing method according to claim 1, characterized in that: During the operation, multiple environmental variables are simulated to verify the game response from the player's perspective. The game stability is evaluated after monitoring the changes in the game interface under various simulated environments through the CV model, including the following steps: Simulate multiple environmental variables during operation, including network latency, screen brightness, battery level, terminal device temperature, storage space, and volume settings; Obtain network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor; Substitute the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor into the linear analysis algorithm to calculate the overall abnormality coefficient of the game under different environmental variables; The overall abnormality coefficient is compared with the preset coefficient threshold. The coefficient threshold is used to evaluate whether the game stability is good or poor. If the overall abnormality coefficient is greater than the coefficient threshold, the game stability is evaluated to be poor. If the overall abnormality coefficient is less than or equal to the coefficient threshold, the game stability is evaluated to be excellent.

3. The AI ​​automated game software testing method according to claim 2, characterized in that: Substitute the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor into the linear analysis algorithm to calculate the overall abnormality coefficient of the game under different environmental variables. The expression is: In the formula, YZ x is the overall abnormal coefficient, {y1, y2, y3, y4, y5, y6} are respectively the network delay operation abnormality factor, screen brightness operation abnormality factor, power level operation abnormality factor, terminal device temperature operation abnormality factor, storage space operation abnormality factor, and volume setting operation abnormality factor. {w1, w2, w3, w4, w5, w6} are the regression coefficients of the abnormal operation factors of each environmental variable, and the regression coefficients are greater than 0.

4. The AI ​​automated game software testing method according to claim 3, characterized in that: The automated testing system parses the test case through the test case configuration table, extracts the content that needs to be tested in the game software, generates the corresponding initial test task according to the test case configuration table, and configures the CV model, including the following steps: The automated test console reads information from the predefined test case configuration table, which records the content of each test case, including the test process, test objectives, and operation steps; Parse the contents of the test case configuration table and extract each test case in the test case configuration table one by one. Each test case contains the steps of the test process, target expected results, test conditions and execution standards; Based on the definition of the test case, the automated test console identifies the test process and test objectives of the test case. The process includes operation steps, and the objective is the expected result of the test. After parsing and extracting the process and target, the automated test console generates the corresponding initial test task, which contains the description of each test step, including the execution order, expected results, and operation type information; According to the process and objectives of each test case, the CV model that needs to be configured is identified, and different models are used to detect different game elements.

5. The AI ​​automated game software testing method according to claim 4, characterized in that: The automated test console sends a startup command to the test assistant APP on the terminal device. The test assistant initializes the test environment according to the initial test task and sends a CV information request to the test assistant APP on the terminal device, including the following steps: The automated test console sends a start command to the test assistant, instructing it to start the test process and prepare for the test. The command includes the test case ID, task information, and CV model requirements required to start the test. After receiving the start command, the test assistant first checks its own status. If any abnormality is found, the test assistant returns a status report and notifies the automated test console to make corrections or restart the command; The test assistant loads the corresponding task configuration according to the initial test task information contained in the startup instruction, and sets up the corresponding test environment according to the requirements of the initial test task. The environment initialization involves simulating network conditions, setting screen resolution, adjusting brightness, and configuring sound. After the environment is prepared, the test assistant sends a CV information request to the client terminal device, and the CV information request includes the identification requirements.

6. The AI ​​automated game software testing method according to claim 5, characterized in that: The test assistant captures the current screen image of the game software, and the CV server recognizes the screen image through multiple CV models to obtain multiple CV information, including the following steps: The test assistant APP runs on the terminal device and captures the screen image of the current game interface according to the request of the automated test console. The image capture includes the entire screen content, so that the automated test system can obtain UI elements, text and scene background information; The test assistant packages the captured screen image data and transmits it to the CV model recognition server through the network. If packet loss or errors occur during the image data transmission process, the test assistant retransmits the image until the server successfully receives it. After receiving the image data, the CV model recognition server loads the CV model according to the test requirements. The CV model includes the target detection model, the text recognition model, the scene classification model, and the instance segmentation model. The object detection model is used to detect UI elements on the screen and identify their position and size; the text recognition model is used to identify text information in the game interface; The scene classification model is used to analyze the background and overall layout of the game scene and identify the characteristics of different scenes; The instance segmentation model is used to segment different areas on the image to distinguish which areas on the image contain content that needs to be further identified, and accurately locate the target area that needs to be detected at the pixel level; Each CV model is called in turn to analyze the screen image, and each CV model extracts information at different levels of the image; The text recognition model extracts the text content on the screen and converts it into a readable data format. The scene classification model identifies the overall layout and background features of the image to determine whether it meets the specified game scene requirements. The test assistant performs fusion analysis on the information obtained from each CV model, packages the CV information obtained from the analysis, and returns it to the automated test console.

7. The AI ​​automated game software testing method according to claim 6, characterized in that: The test assistant performs corresponding operation steps on the game software according to the test case configuration table, including the following steps: The test assistant receives and parses the test case configuration table from the automated test console, and determines the operation steps to be performed according to the task instructions in the test case configuration table; The test assistant uses the recognition results of the target detection model to confirm the position and status of the UI elements that need to be interacted with in the interface, ensuring that the target elements are visible and in an interactive state. Otherwise, an exception is recorded and the operation is skipped. According to the operation type specified in the test case configuration table, the test assistant performs corresponding user interaction behaviors on the terminal device. User interaction behaviors include clicking, sliding, and long pressing to simulate the user's operation habits.

8. An AI automated game software testing system, used to implement the testing method according to any one of claims 1 to 7, characterized in that: Including initialization configuration module, image recognition module, test module, and report generation module; Initialization configuration module: parse the test case through the test case configuration table, extract the content that needs to be tested in the game software, generate the corresponding initial test task according to the test case configuration table, and configure the CV model; Image recognition module: Initializes the test environment according to the initial test task, and sends a CV information request to the test assistant APP on the terminal device. The test assistant captures the current screen image of the game software, recognizes the screen image through multiple CV models, obtains multiple CV information, and determines the game interface status, player operation and environmental feedback after fusion analysis of multiple CV information; Testing module: Execute corresponding operation steps on the game software according to the test case configuration table, simulate multiple environmental variables during the operation, and evaluate the game stability after monitoring the changes of the game interface in various simulated environments through the CV model; Report generation module: After the test is completed, the automated test console automatically generates a test report, which includes game evaluation results, error logs, CV information, operation steps and environment variables.

Images