Video quality evaluation method, device, equipment, medium and product

Abstract

The invention relates to a video quality evaluation method and apparatus, a device, a medium and a product. The method comprises the steps of obtaining a target video, a preset video quality objective evaluation model and a preset subjective evaluation grade classification; according to the subjective evaluation grade classification, determining a first value range when a first model parameter of the video quality objective evaluation model is optimized and a second value range when a second model parameter of the video quality objective evaluation model is optimized; and obtaining subjective evaluation grade results of a plurality of users for the target video, and continuously iteratively correcting the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results until the output precision of the corrected video quality objective evaluation model accords with a preset precision threshold value, and obtaining a target video quality evaluation model to determine a quality score value of the target video. By adopting the method, the accuracy and reliability of objectively evaluating the quality of the target video can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of video processing, and in particular to a video quality evaluation method, device, equipment, medium and product. Background Art

[0002] Quality of Experience (QoE) is a service evaluation method based on user recognition. It integrates the influencing factors at the service level, user level, and environmental level, and directly reflects the user's recognition of the service.

[0003] In the related technologies, the methods for video quality assessment are mainly divided into two methods: subjective quality assessment and objective quality assessment. Subjective quality assessment is to directly let users rate the video quality. This method is the most accurate, but it consumes manpower and time and is not suitable for large-scale application. Objective quality assessment methods use algorithms or models to quantify the video quality, mainly considering the network's throughput, latency, packet loss rate, jitter, bit error rate and other performance. Although it can be automated, the accuracy is usually not as good as subjective evaluation. In other words, the reliability of existing objective video quality evaluation methods is not high. Summary of the invention

[0004] Based on this, it is necessary to address the technical problem of low reliability of the above-mentioned existing objective video quality evaluation methods, consider the idea of ​​combining subjective and objective, and provide a video quality evaluation method, device, computer equipment, computer-readable storage medium and computer program product.

[0005] In a first aspect, the present application provides a video quality evaluation method, the method comprising:

[0006] Obtaining a target video, a preset objective video quality evaluation model, and a preset subjective evaluation grade classification, wherein a first model parameter in the objective video quality evaluation model is used to characterize the degree of influence of a network packet loss rate on video quality, and a first model parameter in the objective video quality evaluation model is used to characterize the degree of correction of the influence of a network packet loss rate on video quality;

[0007] Determine a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determine a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0008] Obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining the target video quality evaluation model;

[0009] According to the target video quality evaluation model, the quality score value of the target video is determined.

[0010] In one embodiment, determining a first value range for optimizing a first model parameter according to the subjective evaluation level classification includes:

[0011] Obtain a mapping relationship table of preset subjective evaluation grade classifications and evaluation scores, and determine the highest evaluation score corresponding to the highest evaluation grade and the lowest evaluation score corresponding to the lowest evaluation grade;

[0012] According to the highest evaluation score, an upper limit value of the first model parameter is determined, and according to the lowest evaluation score, a lower limit value of the first model parameter is determined;

[0013] A first value range is determined according to a lower limit value of the first model parameter and an upper limit value of the first model parameter.

[0014] In one embodiment, determining the second value range for optimizing the second model parameter according to the subjective evaluation level classification includes:

[0015] Obtaining a maximum transmission packet loss rate of a wide area network where the target video is located, and determining an initial value range of a second model parameter according to the maximum transmission packet loss rate;

[0016] The transmission packet loss rate is determined as an independent variable parameter of the objective video quality evaluation model, and sampling is performed within an initial value range to obtain sampling data, wherein the sampling data includes a network packet loss rate value and an evaluation value obtained after the network packet loss rate value is input into the objective video quality evaluation model;

[0017] According to each evaluation value, the upper limit value of the second model parameter is determined, and according to the upper limit value of the second model parameter, the initial value range is updated to obtain the second value range.

[0018] In one embodiment, the first model parameter and the second model parameter are continuously and iteratively corrected according to the first value range and the second value range until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, including:

[0019] Determine, based on the first value range and the second value range, a first initial value of the first model parameter and a second initial value of the second model respectively;

[0020] Iterating the first initial value and the second initial value for a preset number of times with equal step size to obtain a plurality of first quality score output values, and determining an updated first initial value and an updated second initial value according to the first quality score output values;

[0021] The updated first initial value and the updated second initial value are iterated a preset number of times with equal step size to obtain multiple second quality score output values, and the updated first initial value and the updated second initial value are continuously optimized according to the second quality score output values ​​until the output accuracy of the revised video quality objective evaluation model meets the preset accuracy threshold.

[0022] In one embodiment, determining an updated first initial value and an updated second initial value according to the first quality score output value includes:

[0023] Determine a corresponding network damage value according to each first quality score output value, obtain a subjective evaluation value of at least one user according to the network damage value, and determine an average value of the subjective evaluation values;

[0024] Obtain a preset deviation optimization function, input the average values ​​of the one-to-one corresponding network damage values ​​and the subjective evaluation values ​​into the deviation optimization function to obtain the deviation value, determine the first initial value corresponding to the minimum deviation value as the updated first initial value, and determine the second initial value corresponding to the minimum variation value as the updated second initial value.

[0025] In one embodiment, the first model parameter and the second model parameter are continuously and iteratively corrected according to the first value range, the second value range and the subjective evaluation level result until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, including:

[0026] Determine key points and their corresponding actual quality ratings based on the subjective evaluation results. Key points are used to characterize the mutation points of the video quality evaluation level.

[0027] According to the first value range and the second value range, the first model parameter and the second model parameter are continuously iterated and corrected until the deviation between the output value of the corrected objective video quality evaluation model and the actual quality score value of the corresponding key point is no greater than the preset deviation.

[0028] In a second aspect, the present application further provides a video quality evaluation device, the device comprising:

[0029] An evaluation data acquisition module is used to acquire a target video, a preset objective evaluation model for video quality, and a preset subjective evaluation grade classification, wherein the first model parameter in the objective evaluation model for video quality is used to characterize the degree of influence of the network packet loss rate on the video quality, and the first model parameter in the objective evaluation model for video quality is used to characterize the degree of correction of the influence of the network packet loss rate on the video quality;

[0030] A model parameter optimization module, for determining a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determining a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0031] An evaluation model generation module is used to obtain subjective evaluation grade results of multiple users for a target video, and continuously and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining a target video quality evaluation model;

[0032] The video quality evaluation module is used to determine the quality score value of the target video according to the target video quality evaluation model.

[0033] In a third aspect, the present application further provides a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the following steps when executing the computer program:

[0034] Obtaining a target video, a preset objective video quality evaluation model, and a preset subjective evaluation grade classification, wherein a first model parameter in the objective video quality evaluation model is used to characterize the degree of influence of a network packet loss rate on video quality, and a first model parameter in the objective video quality evaluation model is used to characterize the degree of correction of the influence of a network packet loss rate on video quality;

[0035] Determine a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determine a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0036] Obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining the target video quality evaluation model;

[0037] According to the target video quality evaluation model, the quality score value of the target video is determined.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Obtaining a target video, a preset objective video quality evaluation model, and a preset subjective evaluation grade classification, wherein a first model parameter in the objective video quality evaluation model is used to characterize the degree of influence of a network packet loss rate on video quality, and a first model parameter in the objective video quality evaluation model is used to characterize the degree of correction of the influence of a network packet loss rate on video quality;

[0040] Determine a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determine a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0041] Obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining the target video quality evaluation model;

[0042] According to the target video quality evaluation model, the quality score value of the target video is determined.

[0043] In a fifth aspect, the present application further provides a computer program product, the computer program product comprising a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0044] Obtaining a target video, a preset objective video quality evaluation model, and a preset subjective evaluation grade classification, wherein a first model parameter in the objective video quality evaluation model is used to characterize the degree of influence of a network packet loss rate on video quality, and a first model parameter in the objective video quality evaluation model is used to characterize the degree of correction of the influence of a network packet loss rate on video quality;

[0045] Determine a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determine a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0046] Obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining the target video quality evaluation model;

[0047] According to the target video quality evaluation model, the quality score value of the target video is determined.

[0048] The above-mentioned video quality evaluation method, device, computer equipment, storage medium and computer program product, when evaluating the quality of the target video, first obtain the target video, the preset video quality objective evaluation model and the preset subjective evaluation grade classification, the first model parameter in the video quality objective evaluation model is used to characterize the degree of influence of the network packet loss rate on the video quality, and the first model parameter in the video quality objective evaluation model is used to characterize the degree of correction of the influence of the network packet loss rate on the video quality. Then, according to the subjective evaluation grade classification, the first value range when the first model parameter is optimized is determined, and according to the subjective evaluation grade classification, the second value range when the second model parameter is optimized is determined. Then, the subjective evaluation grade results of multiple users for the target video are obtained, and according to the first value range, the second value range and the subjective evaluation grade results, the first model parameter and the second model parameter are continuously iterated and corrected until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, and the target video quality evaluation model is obtained. Finally, according to the target video quality evaluation model, the quality score value of the target video is determined.

[0049] The video quality evaluation method of the present application continuously iteratively corrects the first model parameter and the second model parameter in a preset video quality objective evaluation model through the user's subjective evaluation level result for the target video, so that the output value of the video quality objective evaluation model continuously approaches the user's subjective evaluation level result, and the corrected video quality objective evaluation model when the final output accuracy meets the preset accuracy threshold is determined as the target video quality evaluation model, which helps to improve the accuracy and reliability of objectively evaluating the quality of the target video. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A diagram showing an application environment of a video quality evaluation method in an embodiment;

[0051] Figure 2 is a flowchart of a video quality evaluation method in one embodiment;

[0052] Figure 3 A flowchart for determining a first value range in one embodiment;

[0053] Figure 4 A flowchart for determining a second value range in one embodiment;

[0054] Figure 5 A flowchart of iterating a first initial value and a second initial value in one embodiment;

[0055] Figure 6 A flowchart of determining an updated second initial value in one embodiment;

[0056] Figure 7A flowchart of correcting an objective evaluation model for video quality in one embodiment;

[0057] Figure 8 is a schematic diagram of a video quality system in one embodiment;

[0058] Fig. 9 is a structural block diagram of a video quality evaluation device in one embodiment;

[0059] Fig.10 is an internal structure diagram of a computer device in one embodiment;

[0060] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] Quality of Experience (QoE) is a service evaluation method based on user recognition. It integrates the influencing factors of the service level, user level, and environmental level, and directly reflects the user's recognition of the service. Regarding the quantification of QoE, the "mean opinion score (MOS)" is currently widely used, which divides the subjective feeling of QoE into five levels.

[0063] In the related technologies, the methods for video quality assessment are mainly divided into two methods: subjective quality assessment and objective quality assessment. Subjective quality assessment is to directly let users rate the video quality. This method is the most accurate, but it consumes manpower and time and is not suitable for large-scale application. Objective quality assessment methods use algorithms or models to quantify the video quality, mainly considering the network's throughput, latency, packet loss rate, jitter, bit error rate and other performance. Although it can be automated, the accuracy is usually not as good as subjective evaluation. In other words, the reliability of existing objective video quality evaluation methods is not high.

[0064] In view of the technical problem that the existing objective video quality evaluation methods are not reliable, this application proposes a method that combines subjective and objective video quality evaluation, aiming to provide a practical and reliable specific model result based on the consideration of network packet loss rate, so as to better evaluate network video quality. Specifically, the video quality evaluation method provided in the embodiments of this application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be but is not limited to various personal computers, laptops, smart phones, tablets, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0065] In one embodiment, Figure 2 As shown, this method is applied to Figure 1 The terminal in the example is used for explanation. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0066] Step 202, obtain the target video, the preset objective video quality evaluation model and the preset subjective evaluation grade classification, the first model parameter in the objective video quality evaluation model is used to characterize the influence of the network packet loss rate on the video quality, and the first model parameter in the objective video quality evaluation model is used to characterize the correction degree of the influence of the network packet loss rate on the video quality.

[0067] Generally, video quality is mainly reflected in three aspects: clarity, fluency, and delay. Video clarity mainly reflects the video resolution during the video playback stage, video fluency mainly reflects the freeze during the video playback stage, and video delay is mainly reflected in the video live switching and video on demand start-up stages. The subjective feeling of the user's eyes is sensitive to clarity and fluency, which is also the main factor affecting subjective scoring. In addition, after the video is transmitted over the network, data packet loss will occur, and packet loss often affects the clarity and fluency of the video. According to a large number of experiments, the video playback quality and network packet loss change show a nonlinear relationship and decrease very quickly. In addition, since the QoE value of video quality evaluation is generally a 5-point system (0~5).

[0068] Among them, the target video is used to characterize the video to be evaluated for quality. The objective evaluation model of video quality can be preset according to the type of target video and the evaluation requirements. The preset subjective evaluation level classification is usually five categories, including very good, good, general, poor and very poor, and the corresponding score values ​​can be 5, 4, 3, 2 and 1. The first model parameter and the second model parameter are both undetermined coefficients of the preset objective evaluation model of video quality. The network packet loss rate refers to the proportion of data packets lost during network transmission to the total number of data packets sent. It is usually expressed in percentage (%) or per thousand (‰). The network packet loss rate directly affects the quality of video transmission. A high packet loss rate will cause video freezes, blurs and delays, thereby reducing the user experience.

[0069] For example, the preset subjective evaluation level classification includes five types: very good, good, average, poor and very poor, and the preset video quality objective evaluation model is the following objective evaluation model calculation formula as an example:

[0070] y(x)=5 / [m (n*x) ];

[0071] Where y(x) is used to characterize the model output value; m is used to characterize the first model parameter; and n is used to characterize the second model parameter. In practical applications, the impact of packet loss rate on video quality is usually not a simple linear relationship. On the contrary, as the packet loss rate increases, the rate of decline in video quality may vary. The first model parameter m can more accurately reflect this impact by adjusting the nonlinear relationship in the model. Within a certain range of packet loss rates, the impact of packet loss rate on video quality may have a specific trend of change. The second model parameter n is used to correct this trend of change so that the model can provide accurate evaluation within different packet loss rate ranges. The first model parameter and the second model parameter work together to improve the accuracy and reliability of the final fitting effect.

[0072] Step 204, determining a first value range for optimizing the first model parameter according to the subjective evaluation level classification, and determining a second value range for optimizing the second model parameter according to the subjective evaluation level classification.

[0073] In order to reflect the subjective experience of users and improve the adaptability, accuracy, robustness and reliability of the model, the first value range when optimizing the first model parameters and the second value range when optimizing the second model parameters are determined according to the subjective evaluation level classification.

[0074] Specifically, according to the above content, since the score values ​​corresponding to the preset subjective evaluation level classification in this embodiment are 5, 4, 3, 2 and 1 respectively, in order to ensure that the output value of the final optimized model remains between [1,5], it is necessary to determine the value range of the first model parameter and the value range of the second model parameter.

[0075] Step 206, obtain subjective evaluation level results of multiple users for the target video, and iteratively correct the first model parameters and the second model parameters according to the first value range, the second value range and the subjective evaluation level results until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining the target video quality evaluation model.

[0076] In order to accurately integrate the user's subjective experience and the objective evaluation of the video, and finally obtain a video quality evaluation model with high reliability, it is necessary to continuously iterate and correct the first model parameters and the second model parameters, and in each iteration, adjust the model parameters (the first model parameters and the second model parameters), calculate the error between the model output and the subjective evaluation result, and evaluate the accuracy of the model by calculating the error between the model output and the subjective evaluation result. When the accuracy of the model output meets the preset accuracy threshold, stop the iterative optimization process and obtain the final target video quality evaluation model.

[0077] Step 208: Determine the quality score of the target video according to the target video quality evaluation model.

[0078] After obtaining the target video quality evaluation model, the quality score value of the target video can be determined according to the actual packet loss rate, and the reliability of the quality score value is high.

[0079] In the above-mentioned video quality evaluation method, the first model parameters and the second model parameters in the preset video quality objective evaluation model are continuously iteratively corrected through the user's subjective evaluation level results for the target video, so that the output value of the video quality objective evaluation model continuously approaches the user's subjective evaluation level results, and the corrected video quality objective evaluation model when the final output accuracy meets the preset accuracy threshold is determined as the target video quality evaluation model, which helps to improve the accuracy and reliability of objectively evaluating the quality of the target video.

[0080] In one embodiment, Figure 3 As shown, according to the subjective evaluation level classification, the first value range of the first model parameter optimization is determined, including:

[0081] Step 302: Obtain a mapping relationship table between preset subjective evaluation grade classifications and evaluation scores, and determine the highest evaluation score corresponding to the highest evaluation grade and the lowest evaluation score corresponding to the lowest evaluation grade.

[0082] The preset mapping relationship table between subjective evaluation grade categories and evaluation scores is used to characterize the evaluation scores corresponding to specific subjective evaluation grade categories. According to the above content, in this embodiment, the scoring values ​​(evaluation scores) corresponding to the five subjective evaluation grade categories of very good, good, average, poor and very poor are 5, 4, 3, 2 and 1 respectively, that is, the highest evaluation score is 5 and the lowest evaluation score is 1.

[0083] Specifically, after the highest evaluation score and the lowest evaluation score are determined, the optimization boundaries of the first model parameters and the second model parameters in the iterative optimization process are determined.

[0084] Step 304: determine the upper limit value of the first model parameter according to the highest evaluation score, and determine the lower limit value of the first model parameter according to the lowest evaluation score.

[0085] Specifically, in order to ensure that the output value of the target video quality evaluation model is not greater than 5, the value of the first model parameter m must be greater than 1, so the lower limit of the first model parameter is 1. But at the same time, the value of the first model parameter m cannot be too large, otherwise the output value of the corresponding target video quality evaluation model will be too small. After experiments, it was found that when the maximum value of the first model parameter m reaches 10, the output value of the target video quality evaluation model will fluctuate between 1 and 5, that is, the upper limit of the first model parameter is 10.

[0086] Step 306: Determine a first value range according to the lower limit value of the first model parameter and the upper limit value of the first model parameter.

[0087] Exemplarily, according to step 304, the first value range corresponding to the first model parameter is (1, 10].

[0088] In this embodiment, the value range of the first model parameter is determined based on the user's subjective evaluation results of the target video, which can clarify the user's subjective feelings under different video qualities. After associating these subjective feelings with the model parameters, the user's actual experience can be more accurately captured.

[0089] In one embodiment, Figure 4 As shown, according to the subjective evaluation level classification, the second value range of the second model parameter optimization is determined, including:

[0090] Step 402, obtaining the maximum transmission packet loss rate of the wide area network where the target video is located, and determining the initial value range of the second model parameter according to the maximum transmission packet loss rate.

[0091] Specifically, the maximum transmission packet loss rate reflects the worst transmission condition of the video in the actual network. By obtaining this data, it can be ensured that the model parameters can cover the most extreme network conditions during optimization, thereby improving the robustness and adaptability of the model.

[0092] For example, assuming that the target video is transmitted in a wide area network, after measurement, it is known that the maximum transmission packet loss rate of the local area network where the target video is located is 10%, then it can be determined that the initial value of the second model parameter does not exceed 20, so as to ensure that the optimization process of the model parameter does not exceed the actual network condition. When the packet loss rate is 10%, the preset model calculates the minimum value as y(0.1)=5 / [10 (20*0.1) ], which is 0.05.

[0093] Step 404, determine the transmission packet loss rate as the independent variable parameter of the objective video quality evaluation model, sample within the initial value range to obtain sampled data, the sampled data includes the network packet loss rate value and the evaluation value obtained after the network packet loss rate value is input into the objective video quality evaluation model.

[0094] Specifically, by sampling within the initial value range, a series of network packet loss rates and their corresponding video quality evaluation values ​​can be obtained. These sampled data provide actual training data for the optimization of model parameters, helping the model to make fine adjustments under different network conditions.

[0095] Exemplarily, still taking the example in step 402 as an example, the values ​​of the second model parameters are taken within the initial value range [0,20]. Assuming that the values ​​are 0, 2%, 4%, 6%, 8% and 10% respectively, the corresponding evaluation values ​​are 5.0, 4.8, 4.5, 4.2, 3.8 and 3.5 respectively.

[0096] Step 406, determining the upper limit value of the second model parameter according to each evaluation value, and updating the initial value range according to the upper limit value of the second model parameter to obtain the second value range.

[0097] Specifically, by analyzing the evaluation values ​​in the sampled data, the upper limit of the second model parameter is determined to ensure that the model parameter does not exceed the effective range during the optimization process, avoiding over-optimization or under-optimization of the parameters. In addition, by updating the initial value range, the final second model parameter value range is obtained, providing clear goals and restrictions for the final optimization of the model.

[0098] For example, the maximum transmission packet loss rate of the wide area network where the target video is located is 2%, the initial value of the first model parameter m is 10, and the initial value of the second model parameter is 20. When the input value of the objective video quality evaluation model is 0.02, the score calculated by the objective video quality evaluation model is y(x)=5 / [10 (20*x)]= 1.990535853. Experiments show that the network packet loss rate and the score output by the model are relatively reasonable. Therefore, there is no need to increase the value of n. Therefore, it can be determined that the upper limit value n of the second model parameter can be 20. In other words, the second value range of the second model parameter is (0,20].

[0099] In this embodiment, by obtaining the maximum transmission packet loss rate of the wide area network where the target video is located and determining the initial value range of the second model parameters based on this data, it can be ensured that the model parameters can cover the most extreme network conditions during the optimization process, thereby improving the practical application value, effectiveness and robustness of the model.

[0100] In one embodiment, Figure 5 As shown, according to the first value range and the second value range, the first model parameter and the second model parameter are continuously and iteratively corrected until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, including:

[0101] Step 502: Based on the first value range and the second value range, determine a first initial value of the first model parameter and a second initial value of the second model respectively.

[0102] Specifically, by setting a clear initial value range, we can ensure that the initial values ​​of the model parameters are within a reasonable range, avoiding instability in the optimization process caused by initial values ​​that are too large or too small.

[0103] Exemplarily, taking the value range of the first model parameter m obtained in the above steps as (1, 10] and the value range of the second model parameter n as (0, 20] as an example, the initial value of m is 10 and the initial value of n is 20.

[0104] Step 504, iterate the first initial value and the second initial value for a preset number of times with equal step size to obtain multiple first quality score output values, and determine an updated first initial value and an updated second initial value according to the first quality score output values.

[0105] Specifically, by iterating with equal step length, the parameter space is fully explored, and multiple quality score output values ​​are obtained to provide a data basis for subsequent optimization. That is to say, through multiple iterations, it can help to find the local optimal solution of the model parameters and provide a reference for subsequent optimization. By analyzing the first quality score output value, the optimization direction is clarified, the first initial value is updated, and by analyzing the second quality score output value, the optimization direction is clarified, so that the model parameters are adjusted in a more optimal direction. Among them, the step size and the preset number of times can be preset according to the output accuracy of the target video quality evaluation model finally obtained. The preset number of times in this embodiment can be 100 times as an example.

[0106] For example, the step size of m is 1, the step size of n is 2, and all values ​​are arranged, and the optimization calculation is performed about 100 times, and according to the optimization target MIN T = MIN ∑ [y (X i )-Y i ] 2 , obtain the first round of optimization points m1 (i.e., the first initial value after update) and n1 (i.e., the second initial value after update), where X i is the network packet loss rate; y(X i ) is the model input X i The predicted score value of Y, that is, the first quality score output value; i The user's network packet loss rate is X i When the optimization goal is achieved, that is, the sum of the squares of the difference between the predicted score of the model and the actual score of the user's subjective evaluation is minimized, the updated first initial value m1 and the updated second initial value n1 are obtained.

[0107] Step 506, iterate the updated first initial value and the updated second initial value with equal step size for a preset number of times to obtain multiple second quality score output values, and continue to optimize the updated first initial value and the updated second initial value according to the second quality score output values ​​until the output accuracy of the modified video quality objective evaluation model meets the preset accuracy threshold.

[0108] Specifically, in order to further optimize and obtain the final first model parameters and second model parameters, it is necessary to continue to iterate the updated first initial value and the updated second initial value for a preset number of times based on the updated first initial value (m1) and the updated second initial value (n1) obtained in the first round of optimization to obtain multiple second quality score output values, and continue to optimize the updated first initial value and the updated second initial value according to the second quality score output values, that is, according to the optimization objective MIN T=MIN ∑[y(X i )-Y i ] 2 Continue to optimize until the optimization target reaches the minimum value, that is, the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold. At this time, it can be determined that the first initial value finally updated is the final value of the first model parameter, and the second initial value finally updated is the final value of the second model parameter.

[0109] Exemplarily, based on m1 and n1, the step size is reduced by ten times, and in the second round of optimization, the step size of m is 0.1, and the step size of n is 0.2. The optimization calculation is performed again one hundred times, and the minimum sum of squared differences is taken as the optimization goal. Through the hierarchical approximation optimization algorithm, the y(x) model result is corrected to obtain the final first model parameters and second model parameters, thereby realizing the correction of the objective evaluation model of video quality. The corrected model can be actually applied to video evaluation application scenarios. The set-top box probe obtains the network packet loss value, and after substituting it into the model operation, the quality evaluation value of the video playback can be given.

[0110] In this embodiment, through continuous iterative optimization, it is possible to ensure that the model parameters gradually reach the optimal solution, so that the model output accuracy is gradually improved until it reaches the preset accuracy threshold. This helps to improve the overall performance and reliability of the model and ensure that the model can provide accurate evaluation results under various network conditions.

[0111] In one embodiment, Figure 6 As shown, determining an updated first initial value and an updated second initial value according to the first quality score output value includes:

[0112] Step 602: determine the corresponding network damage value according to each first quality score output value, obtain the subjective evaluation value of at least one user according to the network damage value, and determine the average value of the subjective evaluation value.

[0113] The first quality score output value is the predicted score value of the model, the network damage value is the network packet loss rate value corresponding to the predicted score value of the model, and the user's subjective evaluation value is the subjective evaluation value corresponding to the network damage value.

[0114] Step 604, obtain a preset deviation optimization function, input the average values ​​of the one-to-one corresponding network damage values ​​and the subjective evaluation values ​​into the deviation optimization function to obtain the deviation value, determine the first initial value corresponding to the minimum deviation value as the updated first initial value, and determine the second initial value corresponding to the minimum variation value as the updated second initial value.

[0115] Specifically, the preset deviation optimization function takes the above-mentioned optimization target as an example, and the deviation optimization function in this embodiment is the optimization target: MIN T = MIN ∑ [y (X i )-Y i ] 2 When the deviation optimization function obtains the minimum value, the first initial value corresponding to the minimum deviation value is determined as the updated first initial value, and the second initial value corresponding to the minimum variation value is determined as the updated second initial value.

[0116] In this embodiment, the updated first initial value and the updated second initial value are determined based on the preset deviation optimization function, the network damage value and the average value of the subjective evaluation value, which helps to improve the accuracy of the first initial value and the second initial value when updating.

[0117] In one embodiment, Figure 7 As shown, according to the first value range, the second value range and the subjective evaluation level result, the first model parameter and the second model parameter are continuously and iteratively corrected until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, including:

[0118] Step 702: determine key points and their corresponding actual quality rating values ​​according to the subjective evaluation results. The key points are used to represent the mutation points of the video quality evaluation level.

[0119] Among them, the key point is the point where the user's subjective evaluation score undergoes a mutation.

[0120] Specifically, key points are used to characterize the mutation points of the video quality evaluation level, which helps the model better understand and simulate the key changes in the actual evaluation process. In addition, the subjective evaluation level results reflect the true feelings of humans on video quality. By analyzing these results to determine key points, it can ensure that the model has higher accuracy at key change points, thereby improving the accuracy of the overall evaluation.

[0121] Step 704, continuously and iteratively correct the first model parameter and the second model parameter according to the first value range and the second value range, until the deviation between the output value of the corrected objective video quality evaluation model and the actual quality score value of the corresponding key point is no greater than a preset deviation.

[0122] Specifically, when the deviation between the output value of the corrected objective video quality evaluation model and the actual quality score value at the central key point or its vicinity corresponding to the subjective evaluation is not greater than the preset deviation, for example, the preset deviation in this embodiment is 10%, and the deviation ratio between the output value of the corrected objective video quality evaluation model and the actual quality score value corresponding to the subjective evaluation does not exceed 10%, or after two consecutive rounds of optimization, the deviation ratio is close to 10%, it indicates that the objective video quality evaluation model has been corrected and the result is acceptable.

[0123] For example, in the same playback environment, the subjective rating of the video quality by the personnel is first conducted to correct the objective evaluation model of the video quality. The same playback environment refers to the same set-top box, the same network, and the same video source. The video is played for five minutes, and three viewers observe with their eyes. When they feel "very good, good, average, poor, and very poor", the network control packet loss value is recorded. The network damage meter is experimentally controlled to gradually change the network packet loss, that is, the network packet loss rate is adjusted. There is no packet loss in the initial state.

[0124] The video quality also starts from a very good state. When two of the three viewers feel that the video is no longer "very good", the network damage change is stopped immediately, and the current network damage value X1 is recorded, and the average value of the video quality score of the viewers at this time Y1 is recorded. Continue the above steps and get (Xi, Yi) in sequence, where i∈N * .

[0125] Only one round of subjective video quality scoring is required. The network damage values ​​X1, X2, X3, X4, and X5 are recorded respectively, and the average values ​​of the video quality scores of the viewers are Y1, Y2, Y3, Y4, and Y5.

[0126] The model coefficients m and n are modified and continuously optimized in an attempt to make the objective evaluation model curve approach the key point values ​​and trends of the subjective scoring.

[0127] Optimization goal: MIN T = MIN ∑ [y (Xi) - Yi] 2 =([y(X1)-Y1] 2 +[y(X2)-Y2] 2 +[y(X3)-Y3] 2 +[y(X4)-Y4] 2 +[y(X5)-Y5] 2 ), optimization method: coefficient values ​​are layered and gradually approximated. The initial value of m is 10, and the initial value of n is 20. In the first round, the m step size is 1, and the n step size is 2. All permutations are performed, and the optimization calculation is performed approximately one hundred times to obtain the first round of optimization points m1 and n1. Based on m1 and n1, the step size is reduced by ten times. In the second round, the m step size is 0.1, and the n step size is 0.2. The optimization calculation is performed approximately one hundred times again to obtain the second round of optimization points m2 and n2. This patent only takes two rounds of layered approximation operations in the experiment.

[0128] Finally, with the minimum sum of squared differences as the optimization goal, the y(x) model result is corrected through the hierarchical approximation optimization algorithm. The corrected y(x) model can be actually applied to video evaluation scenarios. The set-top box probe obtains the network packet loss value, and after substituting it into the model calculation, the quality evaluation value of the video playback can be given.

[0129] In this embodiment, the model parameters are modified through continuous iteration to gradually optimize the model performance and make it closer to the optimal solution.

[0130] In a detailed embodiment, Figure 8 As shown in the figure, the objective evaluation model of video quality includes a set-top box player, a network damage meter control, a network packet loss video objective evaluation model and a set-top box probe. The set-top box player is used to play videos, and the observer observes and makes subjective evaluations. The set-top box player controls video playback and displays it on the TV. The network damage meter is connected to the network where the set-top box is located to control the packet loss rate of the set-top box video network. The observer makes subjective evaluations while watching the TV video. The algorithm of the network packet loss video objective evaluation model is integrated in the set-top box probe.

[0131] Determine the objective evaluation model of network packet loss video quality: y(x)=5 / [m (n*x) ], in the experiment, subjective observation of video evaluation, five levels (very good, good, average, poor, very poor), when the video quality level changes suddenly, network damage control subjective observation network damage critical point (Xi, critical packet loss value). X1=0.001 (one thousandth), X2=0.005 (five thousandths), X3=0.01 (one hundredth), X4=0.02 (two percent), X5=0.1 (one tenth). Subjective observation critical point video quality score value (Yi, subjective scoring value). Y1=4.8, Y2=4, Y3=3.2, Y4=2.4, Y5=1.6. The central key point corresponding to the subjective evaluation is (X3, Y3).

[0132] After two rounds of iterative optimization, the optimization results of the model parameters are shown in the following table:

[0133]

[0134] According to the table, the results after model coefficient optimization are: the first model parameter m=6.8, the second model parameter n=17.8, therefore, the actual application model formula of this embodiment is: y(x)=5 / [6.8 (17.8*x) ].

[0135] This embodiment adopts the above method, and continuously iteratively corrects the first model parameter and the second model parameter in the preset video quality objective evaluation model through the user's subjective evaluation level result of the target video, so that the output value of the video quality objective evaluation model continuously approaches the user's subjective evaluation level result, and the corrected video quality objective evaluation model when the final output accuracy meets the preset accuracy threshold is determined as the target video quality evaluation model, which helps to improve the accuracy and reliability of objectively evaluating the quality of the target video.

[0136] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0137] Based on the same inventive concept, the embodiment of the present application also provides a video quality assessment device for implementing the above-mentioned video quality assessment method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the one or more video quality assessment device embodiments provided below can refer to the limitations of the video quality assessment method above, and will not be repeated here.

[0138] In one embodiment, Fig. 9 As shown, a video quality evaluation device is provided, including: an evaluation data acquisition module 902, a model parameter optimization module 904, an evaluation model generation module 906 and a video quality evaluation module 908, wherein:

[0139] Evaluation data acquisition module 902, used to acquire a target video, a preset objective evaluation model for video quality, and a preset subjective evaluation grade classification, wherein the first model parameter in the objective evaluation model for video quality is used to characterize the degree of influence of the network packet loss rate on the video quality, and the first model parameter in the objective evaluation model for video quality is used to characterize the degree of correction of the influence of the network packet loss rate on the video quality;

[0140] A model parameter optimization module 904, for determining a first value range when optimizing a first model parameter according to the subjective evaluation level classification, and determining a second value range when optimizing a second model parameter according to the subjective evaluation level classification;

[0141] An evaluation model generation module 906 is used to obtain subjective evaluation grade results of multiple users for the target video, and continuously and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining a target video quality evaluation model;

[0142] The video quality evaluation module 908 is used to determine the quality score of the target video according to the target video quality evaluation model.

[0143] In one embodiment, the model parameter optimization module 904 is also used to: obtain a mapping relationship table between preset subjective evaluation level classifications and evaluation scores, determine the highest evaluation score corresponding to the highest evaluation level and the lowest evaluation score corresponding to the lowest evaluation level; determine the upper limit value of the first model parameter according to the highest evaluation score, and determine the lower limit value of the first model parameter according to the lowest evaluation score; determine the first value range according to the lower limit value of the first model parameter and the upper limit value of the first model parameter.

[0144] In one embodiment, the model parameter optimization module 904 is also used to: obtain the maximum transmission packet loss rate of the wide area network where the target video is located, and determine the initial value range of the second model parameter according to the maximum transmission packet loss rate; determine the transmission packet loss rate as the independent variable parameter of the objective video quality evaluation model, sample within the initial value range to obtain sampling data, the sampling data includes the network packet loss rate value and the evaluation value obtained after the network packet loss rate value is input into the objective video quality evaluation model; determine the upper limit value of the second model parameter according to each evaluation value, and update the initial value range according to the upper limit value of the second model parameter to obtain the second value range.

[0145] In one embodiment, the evaluation model generation module 906 is also used to: determine the first initial value of the first model parameter and the second initial value of the second model respectively based on the first value range and the second value range; iterate the first initial value and the second initial value with equal step size for a preset number of times to obtain multiple first quality score output values, and determine the updated first initial value and the updated second initial value according to the first quality score output value; iterate the updated first initial value and the updated second initial value with equal step size for a preset number of times to obtain multiple second quality score output values, and continue to optimize the updated first initial value and the updated second initial value according to the second quality score output value until the output accuracy of the modified video quality objective evaluation model meets the preset accuracy threshold.

[0146] In one embodiment, the evaluation model generation module 906 is also used to: determine the corresponding network damage value according to each first quality score output value, obtain the subjective evaluation value of at least one user according to the network damage value, and determine the average value of the subjective evaluation value; obtain a preset deviation optimization function, input the one-to-one corresponding network damage value and the average value of the subjective evaluation value into the deviation optimization function to obtain the deviation value, determine the first initial value corresponding to the minimum deviation value as the updated first initial value, and determine the second initial value corresponding to the minimum variation value as the updated second initial value.

[0147] In one embodiment, the evaluation model generation module 906 is also used to: determine key points and their corresponding actual quality score values ​​based on the subjective evaluation grade results, the key points are used to characterize the mutation points of the video quality evaluation grade; according to the first value range and the second value range, continuously iteratively correct the first model parameters and the second model parameters until the deviation between the output value of the corrected video quality objective evaluation model and the actual quality score value of the corresponding key point is no greater than the preset deviation.

[0148] Each module in the above video quality assessment device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0149] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.10 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store user's subjective evaluation data of the target video, the corresponding network packet loss rate and other data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a video quality evaluation method is implemented.

[0150] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.11As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a video quality evaluation method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0151] Those skilled in the art will understand that Fig.10 and Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0152] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0153] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0154] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0155] 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, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0156] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium 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), magnetoresistive 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. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0157] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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.

[0158] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A video quality evaluation method, characterized in that: The method comprises: Obtaining a target video, a preset objective video quality evaluation model, and a preset subjective evaluation grade classification, wherein a first model parameter in the objective video quality evaluation model is used to characterize the degree of influence of a network packet loss rate on video quality, and a first model parameter in the objective video quality evaluation model is used to characterize the degree of correction of the influence of a network packet loss rate on video quality; Determining a first value range when optimizing the first model parameter according to the subjective evaluation level classification, and determining a second value range when optimizing the second model parameter according to the subjective evaluation level classification; Obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameter and the second model parameter according to the first value range, the second value range, and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining a target video quality evaluation model; According to the target video quality evaluation model, a quality score value of the target video is determined.

2. The method according to claim 1, characterized in that The step of determining a first value range for optimizing the first model parameter according to the subjective evaluation level classification includes: Obtain a preset mapping relationship table of subjective evaluation grade classification and evaluation score, and determine the highest evaluation score corresponding to the highest evaluation grade and the lowest evaluation score corresponding to the lowest evaluation grade; Determine an upper limit value of the first model parameter according to the highest evaluation score, and determine a lower limit value of the first model parameter according to the lowest evaluation score; The first value range is determined according to a lower limit value of the first model parameter and an upper limit value of the first model parameter.

3. The method according to claim 1, characterized in that The step of determining the second value range for optimizing the second model parameter according to the subjective evaluation level classification includes: Obtaining a maximum transmission packet loss rate of a wide area network where the target video is located, and determining an initial value range of the second model parameter according to the maximum transmission packet loss rate; Determine the transmission packet loss rate as the independent variable parameter of the objective video quality evaluation model, perform sampling within the initial value range, and obtain sampling data, wherein the sampling data includes a network packet loss rate value and an evaluation value obtained after the network packet loss rate value is input into the objective video quality evaluation model; According to each of the evaluation values, an upper limit value of the second model parameter is determined, and according to the upper limit value of the second model parameter, the initial value range is updated to obtain the second value range.

4. The method according to claim 1, characterized in that: The step of continuously and iteratively correcting the first model parameter and the second model parameter according to the first value range and the second value range until the output accuracy of the corrected video quality objective evaluation model meets a preset accuracy threshold includes: Based on the first value range and the second value range, respectively determine a first initial value of the first model parameter and a second initial value of the second model; Iterating the first initial value and the second initial value a preset number of times with equal step size to obtain a plurality of first quality score output values, and determining an updated first initial value and an updated second initial value according to the first quality score output values; Iterate the updated first initial value and the updated second initial value a preset number of times with equal step size to obtain multiple second quality score output values, and continue to optimize the updated first initial value and the updated second initial value according to the second quality score output values ​​until the output accuracy of the modified video quality objective evaluation model meets the preset accuracy threshold.

5. The method according to claim 4, characterized in that The step of determining an updated first initial value and an updated second initial value according to the first quality score output value includes: Determine a corresponding network damage value according to each of the first quality score output values, obtain a subjective evaluation value of at least one user according to the network damage value, and determine an average value of the subjective evaluation values; Obtain a preset deviation optimization function, input the one-to-one corresponding average values ​​of the network damage value and the subjective evaluation value into the deviation optimization function to obtain the deviation value, determine the first initial value corresponding to the minimum deviation value as the updated first initial value, and determine the second initial value corresponding to the minimum variation value as the updated second initial value.

6. The method according to claim 1, characterized in that The step of continuously and iteratively correcting the first model parameter and the second model parameter according to the first value range, the second value range, and the subjective evaluation grade result until the output accuracy of the corrected video quality objective evaluation model meets a preset accuracy threshold includes: Determine key points and their corresponding actual quality rating values ​​according to the subjective evaluation results, wherein the key points are used to characterize the mutation points of the video quality evaluation level; According to the first value range and the second value range, the first model parameter and the second model parameter are continuously and iteratively corrected until the deviation between the output value of the corrected objective video quality evaluation model and the actual quality score value of the corresponding key point is no greater than a preset deviation.

7. A video quality evaluation device, characterized in that: The device comprises: An evaluation data acquisition module, used to acquire a target video, a preset video quality objective evaluation model, and a preset subjective evaluation grade classification, wherein the first model parameter in the video quality objective evaluation model is used to characterize the degree of influence of the network packet loss rate on the video quality, and the first model parameter in the video quality objective evaluation model is used to characterize the degree of correction of the influence of the network packet loss rate on the video quality; A model parameter optimization module, used to determine a first value range when optimizing the first model parameter according to the subjective evaluation level classification, and to determine a second value range when optimizing the second model parameter according to the subjective evaluation level classification; An evaluation model generation module is used to obtain subjective evaluation grade results of multiple users for the target video, and iteratively correct the first model parameters and the second model parameters according to the first value range, the second value range and the subjective evaluation grade results, until the output accuracy of the corrected video quality objective evaluation model meets the preset accuracy threshold, thereby obtaining a target video quality evaluation model; The video quality evaluation module is used to determine the quality score value of the target video according to the target video quality evaluation model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images