Video quality evaluation method, device and electronic equipment

Abstract

The present invention provides a video quality evaluation method, device and electronic device, the method comprising: inputting the acquired video to be evaluated into a quality evaluation model pre-trained under a first device, obtaining a first evaluation result when the video to be evaluated is played on the first device; based on the quality evaluation mapping relationship between the second device and the first device and the first evaluation result, determining a second evaluation result when the video to be evaluated is played on the second device. This method uses the quality evaluation model trained under the first device to obtain the quality evaluation result of the video when it is played on the first device, and then obtains the quality evaluation result of the video on other devices through the quality evaluation mapping relationship between different devices. This method does not need to train and run the corresponding quality evaluation model for different devices, thereby reducing the cost of performing video quality evaluation on different devices, and also avoiding the problem of time-consuming model prediction and high resource occupation of low-performance processing devices.

Description

Technical Field

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

[0002] With the development of multimedia technology and display devices, videos are no longer only viewed on PCs (Personal Computers) after being captured, but also on display devices such as iPads and mobile phones. Due to differences in resolution, screen size, etc., the visual perception of the same video on different devices will also be different. Subjective quality evaluation of videos viewed on different viewing devices has become an important way and means to understand users' subjective perceptions.

[0003] In the related art, two methods are provided for evaluating the quality of videos on different devices. One is that the evaluator watches the video on different devices in a certain environment and scores the quality evaluation results on different devices, but this method requires a lot of manpower and financial resources; the other is to score the video through the quality evaluation algorithm, which requires the development and training of corresponding models for different devices, and the use of the model on the corresponding device to predict the quality of unknown quality videos. However, this method develops and trains corresponding subjective quality evaluation models for different devices, which is costly, and the model prediction on machines with weak processing performance such as mobile phones is time-consuming and resource-intensive. Therefore, the technical solutions provided in the related art have problems such as time-consuming and limited processing resources. Summary of the invention

[0004] The object of the present invention is to provide a video quality evaluation method, device and electronic device to improve the cost of cross-platform video quality evaluation and avoid the problem of limited processing resources.

[0005] In a first aspect, the present invention provides a video quality evaluation method, the method comprising: inputting an acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on a first device; wherein the quality evaluation model is trained by a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when played on the first device; based on a preset quality evaluation mapping relationship between a second device and the first device, and the first evaluation result, determining a second evaluation result of the video to be evaluated when it is played on the second device.

[0006] In an optional embodiment, the quality evaluation mapping relationship between the second device and the first device is determined in the following manner: obtaining multiple annotation data; each annotation data includes an annotated video, and a quality evaluation score of the annotated video when played on the first device and a quality evaluation score when played on the second device; performing data fitting on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain the quality evaluation mapping relationship between the second device and the first device.

[0007] In an optional embodiment, the above-mentioned step of performing data fitting on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain the quality evaluation mapping relationship between the second device and the first device includes: substituting the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device into a preset objective function, and determining the numerical value of the function parameter in the objective function; and determining the objective function that determines the function parameter as the quality evaluation mapping relationship between the second device and the first device.

[0008] In an optional embodiment, the above objective function includes:

[0009] Q2 = a / (b+exp(-c*Q1));

[0010] Among them, Q2 represents the quality evaluation score of the annotated video when it is played on the second device; Q1 represents the quality evaluation score of the annotated video when it is played on the first device; a, b and c all represent function parameters.

[0011] In an optional implementation, the quality evaluation model is trained in the following manner: obtaining a sample set; and performing machine learning training on an initial model of the quality evaluation model based on the sample set to obtain a quality evaluation model.

[0012] In an optional embodiment, the above-mentioned step of performing machine learning training on the initial model of the quality evaluation model based on the sample set to obtain the quality evaluation model includes: determining the target sample based on the sample set; inputting the target sample into the initial model to obtain an output result; determining the loss value based on the output result and the quality evaluation score carried by the target sample; adjusting the weight parameters of the initial model according to the loss value, and continuing to perform the step of determining the target sample based on the sample set until the adjusted initial model converges or reaches a preset number of training times to obtain the quality evaluation model.

[0013] In an optional embodiment, the step of determining the second evaluation result of the video to be evaluated when it is played on the second device based on a preset quality evaluation mapping relationship between the second device and the first device, and the first evaluation result, includes: bringing the first evaluation result into the quality evaluation mapping relationship between the second device and the first device, and obtaining the second evaluation result of the video to be evaluated when it is played on the second device.

[0014] In a second aspect, the present invention provides a video quality evaluation device, which includes: a quality evaluation module, used to input the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on a first device; wherein the quality evaluation model is obtained by training a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device; an evaluation conversion module, used to determine a second evaluation result of the video to be evaluated when it is played on a second device based on a preset quality evaluation mapping relationship between the second device and the first device, and the first evaluation result.

[0015] In a third aspect, the present invention provides an electronic device, which includes a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned video quality assessment method.

[0016] In a fourth aspect, the present invention provides a machine-readable storage medium storing machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned video quality assessment method.

[0017] The embodiments of the present invention bring the following beneficial effects:

[0018] The present invention provides a video quality evaluation method, device and electronic device, which first inputs the acquired video to be evaluated into a quality evaluation model pre-trained on a first device to obtain a first evaluation result when the video to be evaluated is played on the first device; then, based on the preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result, determines the second evaluation result when the video to be evaluated is played on the second device. This method uses the quality evaluation model trained on the first device to obtain the quality evaluation result when the video is played on the first device, and then obtains the quality evaluation result of the video on other devices through the quality evaluation mapping relationship between different devices. This method does not need to train and run the corresponding quality evaluation model for different devices, thereby reducing the cost of performing video quality evaluation on different devices, and also avoids the problem of time-consuming model prediction and high resource occupation of low-performance processing devices.

[0019] Other features and advantages of the present invention will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by implementing the above-mentioned technology of the present invention.

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A flowchart of a video quality evaluation method provided by an embodiment of the present invention;

[0023] Figure 2 A flowchart of another video quality evaluation method provided by an embodiment of the present invention;

[0024] Figure 3 A flowchart of a training method for a video quality assessment device provided by an embodiment of the present invention;

[0025] Figure 4 A schematic diagram of the structure of a video quality evaluation device provided by an embodiment of the present invention;

[0026] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0028] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] In the related art, two methods are provided for evaluating the quality of videos played on different devices. One method is to determine the quality through a subjective evaluation test, that is, an evaluator watches the video on different devices in a certain environment and scores the quality evaluation results under different devices. However, this method requires a lot of manpower and financial resources, and requires subjective evaluation experiments to be conducted on each video on different models of devices.

[0030] Another solution is to score videos using a quality evaluation algorithm, which requires developing and training corresponding models for different devices, and using the models on the corresponding devices to predict the quality of the video. To develop and train corresponding subjective quality evaluation models for different devices, on the one hand, it is necessary to subjectively label a large amount of video data for different devices to obtain training data, which is costly. On the other hand, the trained model can be used to make predictions on machines with strong processing performance such as PCs, so real-time performance can be guaranteed. However, model prediction on machines with weaker processing performance such as mobile phones is time-consuming and resource-intensive.

[0031] Based on the above problems, the embodiments of the present invention provide a video quality evaluation method, device and electronic device, which are applied to the scenario of evaluating the quality of videos on different devices. In order to facilitate the understanding of this embodiment, a video quality evaluation method disclosed in the embodiment of the present invention is first described in detail. Figure 1 As shown, the method comprises the following specific steps:

[0032] Step S102, input the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result of the video to be evaluated when it is played on the first device; wherein the quality evaluation model is trained by a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device.

[0033] The video to be evaluated can be a video shot by a camera or a video recorder, or a video in a specified video file, or a video clip captured from a live or recorded video. In specific implementation, the video to be evaluated can be shot by a camera or other device connected by communication and then transmitted, or obtained from a storage device storing the shot video to be processed, or obtained from a storage device storing the specified video file.

[0034] The above-mentioned quality evaluation model can be a neural network model or a deep learning model, etc. The quality evaluation model is usually trained by machine learning using a sample set, which contains a large number of samples, each of which contains a sample video and a quality evaluation score of the sample video when it is played on the first device. The quality evaluation score is the result of the annotator scoring the video playback quality using a unified annotation standard when the sample video is played on the first device. In order to ensure the generalization of the model, the sample videos in the sample set are usually several videos collected in different scenes and different resolutions.

[0035] In a specific implementation, the above-mentioned first device is usually a device with higher resolution and better computing power, for example, a computer or personal computer with a resolution of 1080p or above. The reason why a computer or personal computer with higher resolution is selected as the first device is that the computer or personal computer has stronger computing power, and can better perform model training and quality evaluation result prediction with sufficient computing power, thereby avoiding the problem of high time consumption and resource usage when performing model prediction on machines with weaker processing performance such as mobile phones.

[0036] The video to be evaluated is input into a trained quality evaluation model, and the quality evaluation model can perform quality evaluation on the video to be evaluated, thereby obtaining a first evaluation result when the video to be evaluated is played on a first device, and the first evaluation result is usually a quality evaluation score.

[0037] Step S104: determining a second evaluation result of the video to be evaluated when it is played on the second device based on a preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result.

[0038] The second device is usually a device other than the first device, and the second device can be various models of mobile phones, tablet computers, televisions, etc. The quality evaluation mapping relationship between the second device and the first device is obtained by the R&D personnel in advance based on the quality evaluation scores when the same video is played on the first device and the second device. Specifically, the quality evaluation score of a video played on the first device and the quality evaluation score of a video played on the second device are obtained by annotators in advance, with high accuracy, so that an accurate quality mapping relationship between the first device and the second device can be obtained.

[0039] After obtaining the first evaluation result when the video to be evaluated is played on the first device, the first evaluation result can be converted into a second evaluation result when the video to be evaluated is played on the second device according to the quality evaluation mapping relationship between the second device and the first device.

[0040] A video quality evaluation method provided by an embodiment of the present invention first inputs the acquired video to be evaluated into a quality evaluation model pre-trained on a first device to obtain a first evaluation result when the video to be evaluated is played on the first device; then, based on a preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result, a second evaluation result when the video to be evaluated is played on the second device is determined. This method uses the quality evaluation model trained on the first device to obtain the quality evaluation result when the video is played on the first device, and then the quality evaluation result of the video on other devices can be obtained through the quality evaluation mapping relationship between different devices. This method does not need to train and run the corresponding quality evaluation model for different devices, thereby reducing the cost of performing video quality evaluation on different devices, and also avoiding the problem of time-consuming model prediction and high resource occupation of low-performance processing devices.

[0041] The embodiment of the present invention also provides another video quality evaluation method, which is implemented on the basis of the method of the above embodiment; the method focuses on describing the specific process of determining the quality evaluation mapping relationship between the second device and the first device before obtaining the first evaluation result (implemented by the following steps S202-S204), and the specific process of determining the second evaluation result when the video to be evaluated is played on the second device based on the preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result (implemented by the following step S208); Figure 2 As shown, the method comprises the following specific steps:

[0042] Step S202, obtaining a plurality of annotated data; each annotated data includes an annotated video, and a quality evaluation score of the annotated video when played on a first device and a quality evaluation score of the annotated video when played on a second device.

[0043] The above-mentioned multiple annotated data may be multiple samples extracted from the sample set, and then the sample video in each sample (equivalent to the above-mentioned annotated video) is played on the second device, so that the annotator can perform quality evaluation annotation on the played sample video and obtain the quality evaluation score of the sample video when it is played on the second device, so that each annotated data contains the quality evaluation score of the sample video when it is played on the first device and the quality evaluation score of the sample video when it is played on the second device. In specific implementation, the number of samples extracted from the sample set can be set according to R&D requirements, for example, it can be set to no less than 3.

[0044] In some embodiments, the annotated video in the above-mentioned annotation data may also be a video different from the sample video in the sample set. After obtaining the annotated video, the annotator can use a unified annotation standard to play the annotated video on the first device and the second device respectively, and annotate the playback quality of the annotated video to obtain the quality evaluation score of the annotated video when played on the first device and the quality evaluation score when played on the second device.

[0045] In specific implementation, the range of the quality evaluation scores annotated by the annotators can be set between 0-100 or between 0-50, etc., so as to unify the quality level corresponding to each quality evaluation score.

[0046] Step S204: performing data fitting on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain a quality evaluation mapping relationship between the second device and the first device.

[0047] The quality evaluation score of each video in the plurality of annotated data when played on the first device and the quality evaluation score when played on the second device are used for data fitting, so as to fit the quality evaluation mapping relationship from the first device to the second device. Specifically, the above step S204 can be implemented by the following steps 10-11:

[0048] Step 10: Substitute the quality evaluation score of each labeled video when it is played on the first device and the quality evaluation score of each labeled video when it is played on the second device into a preset objective function to determine the value of the function parameter in the objective function.

[0049] The above objective function can be set according to research and development needs. The unknown function parameters in the objective function are usually constants and can be determined according to the quality evaluation scores of multiple labeled videos when played on the first device and the quality evaluation scores when played on the second device.

[0050] In specific implementation, the above objective function can use the following functions:

[0051] Q2 = a / (b+exp(-c*Q1));

[0052] Among them, Q2 represents the quality evaluation score of the annotated video when it is played on the second device; Q1 represents the quality evaluation score of the annotated video when it is played on the first device; a, b and c all represent function parameters.

[0053] In some embodiments, in order to limit the quality evaluation score when the video is played on the second device, the value range of the quality evaluation score can be constrained by the formula Q2 = max(min(Q2, 100), 0) to limit the value range of the quality evaluation score to between 0 and 100. Specifically, the above formula can also be adjusted to limit the value range of the quality evaluation score to other numerical ranges.

[0054] Step 11: determine the objective function of the function parameter as the quality evaluation mapping relationship between the second device and the first device.

[0055] In a specific implementation, the quality evaluation mapping relationship between the second device and the first device may be represented by a function, thereby more intuitively indicating the quality mapping relationship between different devices.

[0056] Step S206: input the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on a first device.

[0057] Step S208: The first evaluation result is brought into the quality evaluation mapping relationship between the second device and the first device to obtain a second evaluation result when the video to be evaluated is played on the second device.

[0058] The above-mentioned video quality evaluation method, when calculating the video quality evaluation results across devices, only needs to use the quality evaluation model on the first device to predict the quality evaluation results of the video on the first device, and then obtain the quality evaluation scores of the video on other devices through a simple function mapping relationship. This method does not require the development and training of corresponding quality evaluation models for different devices, thereby reducing the cost of predicting video quality evaluation results on different devices.

[0059] The embodiment of the present invention also provides a method for training a quality evaluation model, where the quality evaluation model is a model used to implement the method of the above embodiment; Figure 3 As shown, the training method includes the following specific steps:

[0060] Step S302, obtaining a sample set; each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device.

[0061] The above sample set contains a large number of samples, each of which includes a sample video and a quality evaluation score of the sample video when it is played on the first device. The quality evaluation score is the result of the annotator scoring the video playback quality using a unified annotation standard when the sample video is played on the first device.

[0062] Step S304: Based on the sample set, perform machine learning training on the initial model of the quality evaluation model to obtain a quality evaluation model.

[0063] In specific implementation, the above step S304 can be implemented by the following steps 20-23:

[0064] Step 20: determine the target sample based on the sample set.

[0065] When determining the target sample from the sample set for the first time, a sample is randomly selected from the sample set as the target sample; when determining the target sample pair thereafter, a sample can be randomly selected from the samples in the sample set that have not been determined as the target sample as a new target sample.

[0066] Step 21, input the above target sample into the initial model to obtain the output result.

[0067] The above initial model may be a neural network or a deep learning model, etc. The above output result may be a result obtained after the initial model performs quality evaluation on the sample video in the target sample.

[0068] Step 22, determining the loss value based on the output result and the quality evaluation score carried by the target sample.

[0069] In the specific implementation, the difference between the output result and the quality evaluation score carried by the target sample is determined as the loss value. Specifically, the loss value can be the difference between the output result and the quality evaluation score carried by the target sample, or it can be the difference between the output result and the quality evaluation score carried by the target sample calculated by a preset loss function, and the loss function can adopt a cross entropy loss function, an absolute value loss function, a square loss function, etc.

[0070] Step 23, adjust the weight parameters of the initial model according to the loss value, and continue to perform the step of determining the target sample based on the sample set until the adjusted initial model converges or reaches a preset number of training times to obtain a quality evaluation model.

[0071] In the specific implementation, the back propagation algorithm can be used to find the derivative of the loss value L with respect to all weight parameters W in the initial model. Next, the weight parameter W′ of the initial model is updated by the stochastic gradient descent algorithm: The weight parameters of the initial model are continuously updated iteratively until the initial model converges to obtain a quality evaluation model. Among them, α is the learning rate, which is a hyperparameter pre-set by humans. Common values ​​are 0.01, 0.001, etc.

[0072] The training method of the above-mentioned quality evaluation model, the quality evaluation model trained in this way can accurately evaluate the quality evaluation results of the video when it is played on the first device, thereby improving the accuracy of the subsequent quality evaluation score on the second device obtained based on the quality evaluation mapping relationship between the first device and the second device.

[0073] In view of the above-mentioned quality evaluation method embodiment, the present invention also provides a video quality evaluation device, such as Figure 4 As shown, the device comprises:

[0074] The quality evaluation module 40 is used to input the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on the first device; wherein the quality evaluation model is obtained by training a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device.

[0075] The evaluation conversion module 41 is used to determine a second evaluation result when the video to be evaluated is played on the second device based on a preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result.

[0076] The above-mentioned video quality evaluation device first inputs the acquired video to be evaluated into the quality evaluation model pre-trained under the first device to obtain the first evaluation result when the video to be evaluated is played on the first device; then, based on the preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result, determines the second evaluation result when the video to be evaluated is played on the second device. This method uses the quality evaluation model trained under the first device to obtain the quality evaluation result of the video when it is played on the first device, and then obtains the quality evaluation result of the video on other devices through the quality evaluation mapping relationship between different devices. This method does not need to train and run the corresponding quality evaluation model for different devices, thereby reducing the cost of video quality evaluation on different devices, and also avoids the problem of time-consuming model prediction and high resource occupation of low-performance processing devices.

[0077] Specifically, the above-mentioned device also includes a relationship determination module, which is used to: obtain multiple annotation data; each annotation data includes an annotated video, and a quality evaluation score of the annotated video when played on the first device and a quality evaluation score when played on the second device; perform data fitting on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain a quality evaluation mapping relationship between the second device and the first device.

[0078] Furthermore, the above-mentioned relationship determination module is also used to: bring the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device into a preset objective function to determine the numerical value of the function parameter in the objective function; and determine the objective function that determines the function parameters as the quality evaluation mapping relationship between the second device and the first device.

[0079] In specific implementation, the above objective function includes:

[0080] Q2 = a / (b+exp(-c*Q1));

[0081] Among them, Q2 represents the quality evaluation score of the annotated video when it is played on the second device; Q1 represents the quality evaluation score of the annotated video when it is played on the first device; a, b and c all represent function parameters.

[0082] Specifically, the above-mentioned device also includes a model training module, which is used to: obtain a sample set; based on the sample set, perform machine learning training on the initial model of the quality evaluation model to obtain the quality evaluation model.

[0083] Furthermore, the above-mentioned model training module is also used to: determine the target sample based on the sample set; input the target sample into the initial model to obtain an output result; determine the loss value based on the output result and the quality evaluation score carried by the target sample; adjust the weight parameters of the initial model according to the loss value, and continue to execute the step of determining the target sample based on the sample set until the adjusted initial model converges or reaches a preset number of training times to obtain a quality evaluation model.

[0084] Specifically, the evaluation conversion module 41 is further used to: bring the first evaluation result into the quality evaluation mapping relationship between the second device and the first device to obtain the second evaluation result when the video to be evaluated is played on the second device.

[0085] The video quality assessment device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0086] The embodiment of the present invention further provides an electronic device, such as Figure 5 As shown, the electronic device includes a processor 101 and a memory 100. The memory 100 stores machine executable instructions that can be executed by the processor 101. The processor 101 executes the machine executable instructions to implement the above-mentioned video quality assessment method.

[0087] Further, Figure 5The terminal device shown further includes a bus 102 and a communication interface 103 , and the processor 101 , the communication interface 103 and the memory 100 are connected via the bus 102 .

[0088] The memory 100 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0089] The processor 101 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 101. The above processor 101 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 100, and the processor 101 reads the information in the memory 100 and completes the steps of the method of the above embodiment in combination with its hardware.

[0090] An embodiment of the present invention also provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned video quality evaluation method. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0091] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0092] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A video quality evaluation method, characterized in that: The method comprises: Inputting the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on a first device; wherein the quality evaluation model is obtained by training a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device; Determine, based on a preset quality evaluation mapping relationship between the second device and the first device, and the first evaluation result, a second evaluation result when the video to be evaluated is played on the second device; The quality evaluation mapping relationship between the second device and the first device is determined by: acquiring a plurality of annotated data; each of the annotated data includes an annotated video, and a quality evaluation score of the annotated video when played on the first device and a quality evaluation score of the annotated video when played on the second device; performing data fitting on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain a quality evaluation mapping relationship between the second device and the first device; The step of performing data fitting on the quality evaluation score of each of the annotated videos when played on the first device and the quality evaluation score of each of the annotated videos when played on the second device to obtain a quality evaluation mapping relationship between the second device and the first device comprises: bringing the quality evaluation score of each of the annotated videos when played on the first device and the quality evaluation score of each of the annotated videos when played on the second device into a preset objective function to determine the value of the function parameter in the objective function; determining the objective function for determining the function parameter as the quality evaluation mapping relationship between the second device and the first device; The objective function includes: Q2=a / (b+exp(-c*Q1)); wherein Q2 represents the quality evaluation score of the annotated video when it is played on the second device; Q1 represents the quality evaluation score of the annotated video when it is played on the first device; a, b and c all represent the function parameters.

2. The method according to claim 1, characterized in that The quality evaluation model is trained in the following way: Acquire the sample set; Based on the sample set, machine learning training is performed on the initial model of the quality assessment model to obtain the quality assessment model.

3. The method according to claim 2, characterized in that The step of performing machine learning training on the initial model of the quality evaluation model based on the sample set to obtain the quality evaluation model comprises: Determining a target sample based on the sample set; Inputting the target sample into the initial model to obtain an output result; Determining a loss value based on the output result and the quality evaluation score carried by the target sample; The weight parameters of the initial model are adjusted according to the loss value, and the step of determining the target sample based on the sample set is continued until the adjusted initial model converges or reaches a preset number of training times, thereby obtaining the quality evaluation model.

4. The method according to claim 1, characterized in that The step of determining the second evaluation result when the video to be evaluated is played on the second device based on the preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result comprises: The first evaluation result is brought into the quality evaluation mapping relationship between the second device and the first device to obtain a second evaluation result when the video to be evaluated is played on the second device.

5. A video quality evaluation device, characterized in that: The device comprises: A quality evaluation module, used for inputting the acquired video to be evaluated into a pre-trained quality evaluation model to obtain a first evaluation result when the video to be evaluated is played on a first device; wherein the quality evaluation model is obtained by training a preset sample set, and each sample in the sample set includes a sample video and a quality evaluation score of the sample video when it is played on the first device; An evaluation conversion module, configured to determine a second evaluation result when the video to be evaluated is played on the second device based on a preset quality evaluation mapping relationship between the second device and the first device and the first evaluation result; A relationship determination module, used to obtain multiple annotated data; each of the annotated data includes an annotated video, and a quality evaluation score of the annotated video when played on the first device and a quality evaluation score when played on the second device; data fitting is performed on the quality evaluation score of each annotated video when played on the first device and the quality evaluation score when played on the second device to obtain a quality evaluation mapping relationship between the second device and the first device; The relationship determination module is also used for performing data fitting on the quality evaluation score of each of the annotated videos when played on the first device and the quality evaluation score of each of the annotated videos when played on the second device to obtain the quality evaluation mapping relationship between the second device and the first device, including: bringing the quality evaluation score of each of the annotated videos when played on the first device and the quality evaluation score of each of the annotated videos when played on the second device into a preset objective function to determine the value of the function parameter in the objective function; determining the objective function for determining the function parameter as the quality evaluation mapping relationship between the second device and the first device; The objective function includes: Q2=a / (b+exp(-c*Q1)); wherein Q2 represents the quality evaluation score of the annotated video when it is played on the second device; Q1 represents the quality evaluation score of the annotated video when it is played on the first device; a, b and c all represent the function parameters.

6. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the video quality assessment method according to any one of claims 1 to 4.

7. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the video quality assessment method according to any one of claims 1 to 4.

Images