Video quality assessment methods, devices, electronic equipment and storage media

Abstract

This invention relates to the field of communication technology and discloses a video quality assessment method, comprising: classifying various videos in a video set; inputting videos of different categories into different preset models; and using the preset models to obtain the quality assessment results of the videos. This invention also discloses a video quality assessment device, electronic device, and storage medium. The video quality assessment method, device, electronic device, and storage medium provided by this invention can achieve automated video quality assessment with high efficiency, adaptability to videos in various scenarios, and accurate assessment results for video quality in multiple scenarios.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of communication, in particular to a video quality evaluation method and device, electronic equipment and storage medium. BACKGROUND

[0002] 5G (the fifth generation mobile communication technology) is a new generation of broadband mobile communication technology with high speed, low latency and large connection characteristics, which will lead the world from the mobile Internet era to the mobile Internet of Things era. With the commercialization and popularization of 5G, video services will involve multiple scenarios, such as high-definition video calls, autonomous driving, and remote medical scenarios. In order to timely and accurately grasp the overall running state of the video service system, it is urgent to have an automatic running video quality evaluation system to evaluate the video quality and propose improvement measures for weak links or technical defects of the video quality, so as to continuously improve the running quality of the video service system and meet the increasingly strong quality demands of users on video services.

[0003] Currently, there are mainly two methods for evaluating video quality: one is subjective quality evaluation by evaluators, and the other is objective quality evaluation by establishing mathematical models. However, the former evaluation method is inefficient and difficult to be deployed on a large scale due to the use of manual methods; the latter evaluation method can only adapt to a single scene because the mathematical model is established for videos in a single scene, and therefore it cannot obtain accurate evaluation results for videos in multiple scenes. SUMMARY

[0004] The main purpose of the embodiments of the present application is to provide a video quality evaluation method, device, electronic equipment and storage medium, which can automatically evaluate video quality, is efficient, and can adapt to videos in multiple scenes and obtain accurate evaluation results for video quality in multiple scenes.

[0005] To achieve the above purpose, the embodiments of the present application provide a video quality evaluation method, comprising: classifying each video in a video set; inputting videos of different categories into different preset models to obtain quality evaluation results of the videos by using the preset models.

[0006] To achieve the above purpose, the embodiments of the present application also provide a video quality evaluation device, comprising: an acquisition module configured to classify each video in a video set; and an evaluation module configured to input videos of different categories into different preset models to obtain quality evaluation results of the videos by using the preset models.

[0007] To achieve the above object, the embodiment of the present application further provides an electronic device, comprising: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the video quality evaluation method.

[0008] To achieve the above object, the embodiment of the present application further provides a computer readable storage medium storing a computer program, and the computer program is executed by a processor to implement the video quality evaluation method.

[0009] The video quality evaluation method provided by the present application can realize automatic evaluation of video quality by using preset models to evaluate video quality, and is efficient and suitable for large-scale deployment and application. Meanwhile, different videos are input into different preset models, and different preset models are used to obtain video quality evaluation results, so that videos of different categories in different scenarios can obtain quality evaluation results suitable for the categories, thereby making the video quality evaluation adapt to videos in multiple scenarios and obtaining accurate evaluation results for video quality in multiple scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0010] One or more embodiments are illustrated by way of example with reference to the drawings, which are not intended to be limiting of the embodiments.

[0011] Figure 1 is a flowchart of a video quality evaluation method provided by an embodiment of the present application;

[0012] Figure 2 is a schematic diagram of a principle of weighted sampling of a video quality evaluation method provided by an embodiment of the present application;

[0013] Figure 3 is another flowchart of a video quality evaluation method provided by an embodiment of the present application;

[0014] Figure 4 is a schematic diagram of a principle of measuring a mapping evaluation model in a video quality evaluation method provided by an embodiment of the present application;

[0015] Figure 5 is still another flowchart of a video quality evaluation method provided by an embodiment of the present application;

[0016] Figure 6 is an example diagram of video data collection in a video quality evaluation method provided by an embodiment of the present application;

[0017] Figure 7 is a principle schematic diagram of an end-to-end evaluation model in a video quality evaluation method provided by an embodiment of the present application;

[0018] Figure 8 is a principle schematic diagram of a video quality evaluation method provided by an embodiment of the present application;

[0019] Figure 9 is a training, verification and testing schematic diagram of a preset model in a video quality evaluation method provided by an embodiment of the present application;

[0020] Figure 10 is a module structure schematic diagram of a video quality evaluation device provided by an embodiment of the present application;

[0021] Figure 11 is a structure schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0022] To make the objectives, technical solutions, and superiorities of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below with reference to the drawings. However, those skilled in the art can understand that, in the embodiments of the present application, many technical details are presented in order to make the readers better understand the present application. However, the technical solutions claimed by the present application can be implemented even without these technical details and based on various changes and modifications of the following embodiments. The following embodiments are classified for the convenience of description, and should not constitute any limitation on the specific implementation of the present application. The embodiments can be combined and referenced with each other on the premise of no contradiction.

[0023] In one embodiment, a video quality evaluation method is provided. The method classifies each video in a video set, inputs videos of different categories into different preset models, and uses the different preset models to obtain video quality evaluation results. Since the preset models are used to evaluate video quality, the method can realize automatic evaluation of video quality, has high efficiency, and is suitable for large-scale deployment and application. Meanwhile, by inputting different videos into different preset models and using different preset models to obtain video quality evaluation results, videos of different categories in different scenarios can obtain quality evaluation results that are suitable for the categories, so that the video quality evaluation can adapt to videos in multiple scenarios and accurately evaluate the quality of videos in multiple scenarios.

[0024] It should be noted that the execution subject of the video quality evaluation method provided by the embodiments of the present application can be a server, wherein the server can be implemented by a single server or a cluster composed of multiple servers.

[0025] The specific process of the video quality evaluation method provided by the embodiments of the present application is as followsFigure 1 As shown, the method comprises the following steps:

[0026] S101: classify each video in the video set.

[0027] In one specific example, classifying each video in the video set can be classifying each video in the video set according to at least one of the following: a functional scene, a video length, an access concurrency, an access type, and a network environment parameter, so as to obtain the category of each video. For example, according to the functional scene, the video can be classified into the following categories: a conference, a live broadcast, or an on-demand video, etc.; according to the video length, the video can be classified into the following categories: a long video, a short video, etc. The specific classification method can be determined according to actual needs, and the embodiments of the present application do not make specific limitations thereto.

[0028] S102: input the videos of different categories into different preset models, and obtain the quality evaluation result of the video by using the preset model.

[0029] It can be understood that when different preset models are used, the obtained quality evaluation result of the video is also different. In addition, when the video data of the video at different positions is input into the same preset model, the obtained quality evaluation result of the video can also be different. For example, compared with inputting the video after link transmission into the preset model, the obtained quality evaluation result of the video can be different when the decoded video is input into the preset model. The preset model can be two or more, so as to realize the quality evaluation of the videos of different categories. The specific category and number of the preset model can be determined according to the classification method of each video, and the embodiments of the present application do not make specific limitations thereto.

[0030] Since the number of videos included in the video set can be massive, if the video quality evaluation is performed on each video in the video set, the computing power of the server can be insufficient, and it is also meaningless. In one specific example, before the videos of different categories are input into different preset models, it further comprises the following steps: adding a label and / or a weight to each video in the video set, and extracting part of the videos in the video set by using a weighted sampling algorithm according to the label and / or the weight; and inputting the videos of different categories into different preset models comprises: inputting the videos of different categories in the part of the videos into different preset models. The principle diagram of the weighted sampling of the video can be referred to Figure 2 .

[0031] When adding labels and / or weights to each video in the video set, the video can be added with labels and / or weights according to access concurrency, network environment, and usage scale, and the like. For example, if the video is in a 5G network environment, a label of 5G is added, and if the video is in a 4G network environment, a label of 4G is added. For another example, if the access concurrency is high, a higher weight is added, and if the access concurrency is low, a lower weight is added. The specific manner of adding labels and weights can be set according to actual needs, and the embodiments of the present application do not make specific limitations thereon. After adding labels and / or weights to each video, the weighted sampling algorithm is sampled for each video, so that the videos in the video set with high weights are sampled more and the videos with low weights are sampled less, and representative videos are extracted for evaluation, which can reduce the system pressure brought by massive data.

[0032] By adding labels and / or weights to the video, part of the videos in the video set are extracted according to the labels and / or weights using the weighted sampling algorithm, and the video quality of the extracted part of the videos is evaluated, which can extract representative videos for quality evaluation, can better reflect the overall video quality of the video set, and at the same time, can reduce the burden of the server by reducing the dimensionality of the massive video set.

[0033] The video quality evaluation method provided by the embodiments of the present application can obtain the categories of each video in the video set, input the videos of different categories into different preset models, and obtain the quality evaluation results of the videos by using different preset models. Since the preset model is used to evaluate the video quality, the video quality can be automatically evaluated, the efficiency is high, and the method is suitable for large-scale deployment and application. At the same time, by obtaining the categories of the videos, inputting different videos into different preset models, and obtaining the quality evaluation results of the videos by using different preset models, the videos of different categories in different scenes can obtain quality evaluation results that are adapted to the categories, so that the video quality evaluation can be adapted to videos in multiple scenes, and accurate evaluation results can be obtained for the video quality in multiple scenes.

[0034] In a specific example, the videos in the video set include first category videos, the preset models include a metric mapping evaluation model, and before the different categories of videos are input into different preset models (S102), the video quality evaluation method provided by the embodiments of the present application further includes: obtaining transmission characteristic data of the videos on a video link; and inputting the transmission characteristic data of the first category videos into the metric mapping evaluation model, obtaining a first score of the first category videos by using the metric mapping evaluation model, and the first score is output by the metric mapping evaluation model after evaluation according to the transmission characteristic data.

[0035] Please refer to Figure 3This is another flowchart illustrating the video quality assessment method provided in this embodiment of the invention, specifically including the following steps:

[0036] S101': Classify the videos in the video collection.

[0037] S102': Obtain the transmission characteristic data of the video on the video link.

[0038] Transmission characteristic data refers to the characteristic data of video transmission on the video link, such as packet loss rate, frame loss rate, latency, or jitter.

[0039] To ensure that the trained metric mapping evaluation model can perform quality assessments on all videos, transport feature data unique to each video can be selected during training. Furthermore, the more transport feature data input during training, the more accurate the quality assessment results obtained by the metric mapping evaluation model.

[0040] S103': Input the transmission feature data of the first category of videos into the metric mapping evaluation model, use the metric mapping evaluation model to obtain the first score of the first category of videos, and output the first score after evaluation by the metric mapping evaluation model based on the transmission feature data.

[0041] In practical implementations, the metric mapping evaluation model, such as Figure 4 As shown, KPIs (Key Performance Indicators) can be obtained from video monitoring data (e.g., logs). These KPIs are then used as transmission feature data, mapped to KQIs (Key Quality Indicators), and finally mapped to VMOS (Video Mean Opinion Score). VMOS is used as the first score output by the metric mapping evaluation model. Through this mapping, if the VMOS is low, for example, below a certain score, abnormal transmission feature data can be sent back in a reverse-engineering manner. This allows for the identification of problems causing video quality degradation while simultaneously obtaining video quality assessment results, facilitating video quality improvement.

[0042] It should be understood that the evaluation of the video quality is generally performed by comparing the video watched by the user with the original video, and determining the quality of the video by comparing the difference between the original video and the watched video. However, in some cases, if the original video is difficult to obtain, it is difficult to determine the quality of the video by comparing the difference between the original video and the watched video. For example, in a weak network environment, it is difficult to obtain the original video. At this time, the quality of the video can be evaluated by obtaining the transmission characteristic data of the video link, and obtaining the evaluation score of the quality of the video according to the transmission characteristic data. Accordingly, the video in the weak network environment can be divided into the first category video.

[0043] In a specific example, after obtaining the first score of the first category video by using the metric mapping evaluation model, if the first score is less than the first expected score, the abnormal transmission characteristic data of the video on the video link is located according to the metric mapping evaluation model, and / or the early warning information of the quality of the video is output according to the first score.

[0044] The first expected score can be set according to actual needs, which is not limited here. When the abnormal transmission characteristic data of the video on the video link is located according to the metric mapping evaluation model, the VMOS can be reversed to the KQI, and then the KQI can be reversed to the KPI, and finally the abnormal transmission characteristic data can be located according to the KPI for corresponding abnormal location.

[0045] By locating the abnormal transmission characteristic data of the video on the video link according to the metric mapping evaluation model when the first score is less than the first expected score, the specific problem can be located when the quality of the video is poor, so that targeted measures can be taken for improvement. According to the early warning information of the quality of the video output according to the first score, the user of the video can be informed in advance that the quality of the video will be poor, the user experience can be improved, and the prior prediction of the quality of the video can be realized.

[0046] In a specific example, the video in the video set further includes a second category video, and the preset model further includes an end-to-end evaluation model. Before the videos of different categories are input into different preset models (S102), a collection point is further arranged at the front end and the rear end of the video link of the video, and the front end video data of the video at the front end and the rear end video data at the rear end are collected through the collection point. The videos of different categories are input into different preset models, and the quality evaluation result of the video is obtained by using the preset model (S102). The front end video data and the rear end video data of the second category video are input into the end-to-end evaluation model, and the second score of the second category video is obtained by using the end-to-end evaluation model, wherein the second score is output by the end-to-end evaluation model after comparing the difference between the rear end video data and the front end video data.

[0047] Please refer toFigure 5 FIG. 2 is another flowchart of a method for evaluating video quality according to an embodiment of the present application, which specifically includes the following steps:

[0048] S101”:Classify each video in the video set.

[0049] S102”:Set at least one collection point at the front end and the rear end of the video link of the video, and collect front-end video data at the front end and rear-end video data at the rear end of the video through the collection point.

[0050] In a specific example, when collecting the front-end video data or the rear-end video data of the video through the collection point, the front-end video data and the rear-end video data are collected through bypass copying at the collection point, so as not to affect the normal video link process and not to generate additional burden, realizing user non-perception.

[0051] Please refer to Figure 6 FIG. 3 is an example diagram of collecting video data in the method for evaluating video quality according to an embodiment of the present application. As shown in FIG. 3, the collection point can be set before encoding, before transmission, after transmission, and after encoding of the video link, and the front-end video data and the rear-end video data are obtained through bypass copying. Figure 6

[0052] S103”:Input the front-end video data and the rear-end video data of the second category video into the end-to-end evaluation model, and obtain a second score of the second category video by using the end-to-end evaluation model, the second score being output by the end-to-end evaluation model after comparing the difference between the rear-end video data and the front-end video data.

[0053] When constructing the end-to-end evaluation model, a plurality of full-reference algorithms can be constructed by using machine learning, deep learning, etc., for example, a set of end-to-end evaluation algorithms is formed based on PSNR, VMAF, DVQA, etc., and finally the end-to-end evaluation model is obtained after training. When selecting the basic network, the service power and quality requirements can be determined. For example, VMAF can be selected as the basic network for the link before and after encoding, which can save the service power; for example, DVQA can be selected as the basic network for the link at the generation end and the playback end of the video, which can accurately extract the spatio-temporal joint features of the video.

[0054] Please refer to Figure 7 ​which is a schematic diagram of the principle of the end-to-end evaluation model in the video quality evaluation method provided by the embodiment of the application. The distorted video in the figure is the back-end video data, and the reference video is the front-end video data. The reference video and the distorted video should be of the same type and in the same time period, and they are relative concepts, which can be before and after encoding and decoding, or before and after encoding, as long as there is a difference (loss) between them, which can be used for comparison. That is, the front-end video data and the back-end video data can respectively refer to the video data before encoding and the video data after decoding, or the video data before encoding and the video data after encoding.

[0055] It should be understood that in the case that the original video can be obtained in a good network environment, the quality evaluation can be performed by comparing the original video before and after transmission, and therefore, compared with the condition of the first type of video division, the video in a non-weak network environment can be divided into the second type of video.

[0056] The quality evaluation of the second type of video can be realized by setting the collection points at the front end and the back end of the video link to collect the video data, and then inputting the collected front-end video data and back-end video data into the end-to-end evaluation model to obtain the quality evaluation result of the second type of video.

[0057] In a specific example, after obtaining the second score of the second type of video by using the end-to-end evaluation model (S103'), it further includes: if the second score is lower than the second expected score, inputting the transmission feature data of the second type of video into the metric mapping evaluation model, obtaining the first score of the second type of video by using the metric mapping evaluation model, and / or outputting the early warning information of the video quality according to the second score.

[0058] The second expected score can be set according to actual needs, and the embodiment of the application does not make specific limitations thereto. It should be understood that when the transmission feature data of the video on the video link is obtained (S102'), the transmission feature data of all videos (all videos in the sampling if it is sampling) in the video is obtained, including the first type of video and the second type of video, and therefore the transmission feature data of the second type of video can be directly inputted into the metric mapping evaluation model to obtain the first score. It can be continuously referred to Figure 6 When the video data is collected by the collection point, the transmission feature data (i.e. the metric data in the figure) of the video is also collected.

[0059] By inputting the transmission feature data of the second category video into the metric mapping evaluation model when the score is lower than expected, the first score of the second category video is obtained by using the metric mapping evaluation model, which can further evaluate the quality of the video from two dimensions; at the same time, since the metric mapping model obtains the score through the transmission feature data, the abnormal transmission feature data can be further deduced according to the score, so as to realize the positioning of the video quality problem; in addition, the early warning information of the video quality is output according to the second score, which can make the user of the video know the information that the video quality will be poor in advance, improve the user experience, and realize the prior prediction of the video quality.

[0060] Please refer to Figure 8 , which is a principle example diagram of the video quality evaluation method provided by the embodiment of the application. Specifically, the videos in the video service are uniformly classified, and part of the videos are extracted after weighted sampling. Then, the link collection data of the extracted part of the videos is collected. The metric mapping model is used for video quality evaluation for the first category video, and the end-to-end evaluation model is used for video quality evaluation for the second category video. If the score obtained by using the end-to-end evaluation model is lower than expected, the metric mapping evaluation model can be further used for evaluation, and the abnormal transmission feature data can be further analyzed while another dimension is obtained, so as to realize the prediction and positioning of the corresponding video quality.

[0061] Please refer to Figure 9 , which is an example diagram of pre-model training, verification and testing in the video quality evaluation method provided by the embodiment of the application. Specifically, the pre-model can be trained by collecting corresponding data from the video set respectively. After training to a certain extent, the accuracy of the model is verified by collecting corresponding data. After the model reaches a certain accuracy, the pre-model (i.e. the quality evaluation model in the figure) is applied to testing, so as to obtain the corresponding quality evaluation result (such as the scores of the 98, 32 and 76 videos in the figure).

[0062] In addition, those skilled in the art can understand that the step division of the above various methods is only for the purpose of clear description, and when implemented, one step can be combined or some steps can be split and decomposed into multiple steps, as long as the same logical relationship is included, all are within the protection scope of the patent; adding irrelevant modifications or introducing irrelevant designs in the algorithm or process, but not changing the core design of the algorithm and process are within the protection scope of the patent.

[0063] In one embodiment, a video quality evaluation device 200 is involved, as shown in Figure 10 , which includes an acquisition module 201 and an evaluation module 202, and the functions of each module are described in detail as follows:

[0064] The acquisition module 201 is configured to classify each video in the video set.

[0065] The evaluation module 202 is configured to input the videos of different categories into different preset models to obtain quality evaluation results of the videos by using the preset models.

[0066] Further, the videos include first category videos, and the preset models include a metric mapping evaluation model; the video quality evaluation device 200 provided in the embodiment of the present application further includes a first acquisition module, wherein the first acquisition module is configured to acquire transmission characteristic data of the videos on a video link; and the evaluation module 202 is further configured to input the transmission characteristic data of the first category videos into the metric mapping evaluation model, and obtain a first score of the first category videos by using the metric mapping evaluation model, wherein the first score is output by the metric mapping evaluation model after evaluation according to the transmission characteristic data.

[0067] Further, the video quality evaluation device 200 provided in the embodiment of the present application further includes an evaluation processing module, which is configured to, when the first score is less than a first expected score, backtrace and locate abnormal transmission characteristic data of the videos on the video link according to the metric mapping evaluation model, and / or output early warning information of the quality of the videos according to the first score.

[0068] Further, the videos further include second category videos, and the preset models further include an end-to-end evaluation model; the video quality evaluation device 200 provided in the embodiment of the present application further includes a second acquisition module, wherein the second acquisition module is configured to set at least one acquisition point at a front end and a rear end of the video link of the videos respectively, acquire front end video data at the front end and rear end video data at the rear end of the videos by using the acquisition points, and the evaluation module 202 is further configured to input the front end video data and the rear end video data of the second category videos into the end-to-end evaluation model, and obtain a second score of the second category videos by using the end-to-end evaluation model, wherein the second score is output by the end-to-end evaluation model after comparison between the difference between the rear end video data and the front end video data.

[0069] Further, the video quality evaluation device 200 provided in the embodiment of the present application further includes a reevaluation module, wherein the reevaluation module is configured to, if the second score is lower than a second expected score, input transmission characteristic data of the second category videos into the metric mapping evaluation model, obtain a first score of the second category videos by using the metric mapping evaluation model, and / or output early warning information of the quality of the videos according to the second score.

[0070] Further, the second acquisition module is further configured to acquire the front end video data and the rear end video data by using a bypass copying mode at the acquisition points.

[0071] Further, the acquisition module 201 is further configured to classify each video in the video set according to at least one of the following: a function scene, a video length, an access concurrency, an access type, and a network environment parameter.

[0072] Further, the video quality evaluation device 200 provided by the embodiment of the present application further comprises an extraction module, wherein the extraction module is configured to: add a label and / or a weight to each video in the video set; and extract part of the videos in the video set by using a weighted sampling algorithm according to the label and / or the weight; and the evaluation module 202 is further configured to: input the videos of different categories in the part of the videos into different preset models.

[0073] It can be found that the embodiment corresponds to the device embodiment of the foregoing method embodiment, and the embodiment can be implemented in cooperation with the foregoing method embodiment. The related technical details mentioned in the foregoing method embodiment are still valid in the embodiment. In order to reduce repetition, they will not be described here. Correspondingly, the related technical details mentioned in the embodiment can also be applied to the foregoing method embodiment.

[0074] It is worth mentioning that each module involved in the embodiment is a logical module. In actual application, one logical unit can be one physical unit, or a part of one physical unit, or realized by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present application, units not closely related to solving the technical problems proposed by the present application are not introduced in the embodiment, but this does not mean that there are no other units in the embodiment.

[0075] In one embodiment, an electronic device is provided, such as Figure 11 As shown in the figure, the electronic device comprises: at least one processor 301; and a memory 302 connected with the at least one processor 301; wherein the memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to perform the video quality evaluation method described above.

[0076] The memory and the processor are connected in a bus mode, the bus can include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage stabilizers and power management circuits together, which are well known in the art, and therefore, they will not be further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements such as multiple receivers and transmitters, which provide a unit for communicating with various other devices on the transmission medium. The data processed by the processor is transmitted on the wireless medium through the antenna, further, the antenna also receives data and transmits the data to the processor.

[0077] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. The memory can be used to store data used by the processor in executing operations.

[0078] In one embodiment, a computer readable storage medium storing a computer program is provided. The computer program, when executed by a processor, implements the method embodiments described above.

[0079] That is, a person skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by a program instructing related hardware, the program is stored in a storage medium, and includes a plurality of instructions for causing an apparatus (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0080] A person of ordinary skill in the art can understand that the above-mentioned embodiments are specific embodiments for implementing the present application, and in actual applications, various changes can be made in form and details without departing from the spirit and scope of the present application.

Claims

1. A method of video quality assessment, characterized by, Comprise: According to at least one of the function scene, the video length, the access concurrency, the access type and the network environment parameter, the video set is classified into different categories; Different categories of videos are input into different preset models, and the quality evaluation result of the video is obtained by using the preset model; Among them, the video includes the first category video, the video in the weak network environment is divided into the first category video, and the preset model includes the metric mapping evaluation model; Before the different categories of videos are input into different preset models, it further comprises: Obtain the transmission characteristic data of the video on the video link; the transmission characteristic data is used to indicate the transmission related characteristic data of the video on the video link; The different categories of videos are input into different preset models, and the quality evaluation result of the video is obtained by using the preset model, comprising: The transmission characteristic data of the first category video is input into the metric mapping evaluation model, and the first score of the first category video is obtained by using the metric mapping evaluation model, wherein the first score is obtained by mapping the transmission characteristic data to the key quality index by the metric mapping evaluation model, and then mapping the key quality index to the video average subjective opinion score; The video further includes the second category video, and the video in the non-weak network environment is divided into the second category video, and the preset model further includes an end-to-end evaluation model; Before the different categories of videos are input into different preset models, it further comprises: At least one collection point is arranged at the front end and the rear end of the video link of the video; The front end video data of the video at the front end and the rear end video data at the rear end are collected through the collection point; The different categories of videos are input into different preset models, and the quality evaluation result of the video is obtained by using the preset model, comprising: The front end video data and the rear end video data of the second category video are input into the end-to-end evaluation model, and the second score of the second category video is obtained by using the end-to-end evaluation model, wherein the second score is output by the end-to-end evaluation model after comparing the difference between the rear end video data and the front end video data; the end-to-end evaluation model is obtained by training an end-to-end evaluation algorithm set, and the end-to-end evaluation algorithm set includes PSNR, VMAF and DVQA.

2. The video quality assessment method of claim 1, wherein, After the first score of the first category video is obtained by using the metric mapping evaluation model, it further comprises: If the first score is less than the first expected score, the abnormal transmission characteristic data of the video on the video link is backtracked and located according to the metric mapping evaluation model, and / or the early warning information of the video quality is output according to the first score.

3. The method of claim 1, wherein, After the second score of the second category video is obtained by using the end-to-end evaluation model, it further comprises: If the second score is lower than a second expected score, transmission feature data of the second category video is input into the metric mapping evaluation model, a first score of the second category video is obtained by using the metric mapping evaluation model, and / or early warning information of video quality is output according to the second score.

4. The method of claim 1, wherein, The front-end video data and the back-end video data of the video are collected at the collection point, and specifically: The front-end video data and the back-end video data are collected at the collection point by means of bypass copying.

5. The method of video quality assessment according to claim 1, wherein, Before the different categories of videos are input into different preset models, the method further includes: adding labels and / or weights to each video in the video set; extracting part of the videos in the video set by using a weighted sampling algorithm according to the labels and / or the weights; the step of inputting the different categories of videos into different preset models includes: inputting the different categories of videos in the part of the videos into different preset models.

6. A video quality assessment apparatus characterized by comprising: The method includes: a obtaining module, configured to classify each video in a video set according to at least one of the following: a functional scene, a video length, an access concurrency, an access type, and a network environment parameter; an evaluation module, configured to input different categories of videos into different preset models, and obtain quality evaluation results of the videos by using the preset models; wherein the videos include first category videos, the videos in a weak network environment are divided into the first category videos, and the preset models include a metric mapping evaluation model; The video quality evaluation device further includes a first collection module. The first collection module is configured to obtain transmission feature data of a video on a video link. The evaluation module is further configured to: input the transmission feature data of the first category videos into the metric mapping evaluation model, and obtain a first score of the first category videos by using the metric mapping evaluation model, wherein the first score is obtained by mapping the transmission feature data to a key quality indicator by the metric mapping evaluation model, and then mapping the key quality indicator to a video average subjective opinion score; The videos further include second category videos, the videos in a non-weak network environment are divided into the second category videos, and the preset models further include an end-to-end evaluation model; The video quality evaluation device further includes a second collection module. The second collection module is configured to set at least one collection point at a front end and a back end of a video link of a video, and collect front-end video data of the video at the front end and back-end video data of the video at the back end through the collection point; The evaluation module is further configured to: input the front-end video data and the back-end video data of the second category videos into the end-to-end evaluation model, and obtain a second score of the second category videos by using the end-to-end evaluation model, wherein the second score is output by the end-to-end evaluation model after comparing the difference between the back-end video data and the front-end video data; The end-to-end evaluation model is obtained by training an end-to-end evaluation algorithm set, and the end-to-end evaluation algorithm set includes PSNR, VMAF, and DVQA.

7. An electronic device, comprising: The method includes: at least one processor; and a memory connected in communication with the at least one processor; wherein The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the video quality assessment method according to any one of claims 1 to 5.

8. A computer readable storage medium storing a computer program, characterized in that, The computer program, when executed by a processor, performs the video quality assessment method according to any one of claims 1 to 5.

Images