Method and apparatus for evaluating network quality, edge computing device and medium

Abstract

The application discloses a network quality evaluation method and device, an edge computing device and a medium. The method comprises the following steps: obtaining a game time delay screenshot of a game process of a PC terminal, determining a region of interest of the game time delay screenshot; identifying character information of the region of interest; determining a game time delay of the game process according to the character information, and sending the game time delay to the PC terminal. The game time delay of the PC terminal is accurately determined, and the efficiency of network testing is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and particularly relates to a network quality evaluation method and device, an edge computing device and a medium. BACKGROUND

[0002] At present, game testing work mainly relies on a limited number of optimization personnel to carry out, and there are problems such as small test coverage, low frequency, single environment, etc. Artificial testing mainly tests the game experience and actual network perception by continuously playing the game, and artificially judges the network delay status. The scheme can obtain the real experience of the game, but it consumes more manpower, has small test coverage and low test efficiency. SUMMARY

[0003] The main purpose of the present application is to provide a network quality evaluation method and device, an edge computing device and a medium, which aims to solve the problem of low test efficiency of network delay.

[0004] To achieve the above purpose, the present application provides a network quality evaluation method, which comprises the following steps:

[0005] Obtaining a game delay screenshot of a game process of a PC terminal, and determining a region of interest of the game delay screenshot;

[0006] Recognizing character information of the region of interest;

[0007] Determining a game delay of the game process according to the character information, and sending the game delay to the PC terminal.

[0008] In an embodiment, the step of recognizing the character information of the region of interest comprises:

[0009] Determining a text box in the region of interest and text box information corresponding to the text box, wherein the text box information at least comprises coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box;

[0010] Determining the text box with the first confidence degree greater than a preset first threshold value as a target text box;

[0011] Inputting the target text box into a preset text recognition model to obtain the character information corresponding to the target text box and a second confidence degree corresponding to the character information.

[0012] In an embodiment, the character recognition model comprises a convolution layer, a recurrent layer and a transcription layer, the convolution layer is configured to extract a feature sequence of the target character box; the recurrent layer is configured to determine a label sequence corresponding to the feature sequence; and the transcription layer is configured to output character information of the target character box and the second confidence according to the label sequence.

[0013] In an embodiment, the step of determining the game delay of the game process according to the character information comprises:

[0014] determining the character information with the second confidence greater than a preset second threshold as target character information;

[0015] if the target character information comprises a preset character, obtaining first coordinate information of a character box corresponding to the target character information;

[0016] obtaining second coordinate information of other character boxes in the region of interest;

[0017] determining the game delay according to the first coordinate information and the second coordinate information.

[0018] In an embodiment, the step of determining the game delay according to the first coordinate information and the second coordinate information comprises:

[0019] determining absolute values of distances between the first coordinate information and each of the second coordinate information;

[0020] determining the character information of the character box corresponding to the minimum value in the absolute values as the game delay.

[0021] To achieve the above object, the present application further provides a network quality evaluation device, which comprises:

[0022] a first obtaining module configured to obtain a game delay screenshot of a game process of a PC terminal and determine a region of interest of the game delay screenshot;

[0023] a recognition module configured to recognize character information of the region of interest;

[0024] a determination module configured to determine a game delay of the game process according to the character information and send the game delay to the PC terminal.

[0025] To achieve the above object, the present application further provides a game perception platform, which comprises a memory, a processor and a network quality evaluation program stored in the memory and executable on the processor, and the network quality evaluation program implements each step of the network quality evaluation method when executed by the processor.

[0026] To achieve the above object, the application further provides a network quality evaluation method applied to a game perception platform, which comprises the following steps:

[0027] acquiring a game delay of a game process of a PC terminal in a preset time period and determining an average delay of the game delay in the preset time period;

[0028] acquiring probing data of network quality of the PC terminal, wherein the probing data comprises network jitter and packet loss;

[0029] determining network quality according to the average delay and the probing data.

[0030] In an embodiment, the step of determining network quality according to the average delay and the probing data comprises:

[0031] determining a network type corresponding to the network and acquiring corresponding quality difference conditions according to the network type;

[0032] when the average delay and the probing data satisfy the quality difference conditions, determining that the network is of poor quality;

[0033] when the average delay and the probing data do not satisfy the quality difference conditions, determining that the network is normal.

[0034] To achieve the above object, the application further provides a network quality evaluation device, which comprises:

[0035] a calculation module, configured to acquire a game delay of a game process of a PC terminal in a preset time period and determine an average delay of the game delay in the preset time period;

[0036] a second acquisition module, configured to acquire probing data of network quality of the PC terminal, wherein the probing data comprises network jitter and packet loss;

[0037] an evaluation module, configured to determine network quality according to the average delay and the probing data.

[0038] To achieve the above object, the application further provides an edge computing device, which comprises a memory, a processor and a network quality evaluation program stored in the memory and executable on the processor, wherein the network quality evaluation program implements each step of the network quality evaluation method when executed by the processor.

[0039] To achieve the above object, the application further provides a computer readable storage medium, which stores a network quality evaluation program, wherein the network quality evaluation program implements each step of the network quality evaluation method when executed by a processor.

[0040] The application provides a network quality evaluation method and device, an edge computing device and a medium. A game time delay screenshot of a game process sent by a PC terminal is acquired, and a region of interest of the game time delay screenshot is determined. Character information of the region of interest is identified. The game time delay of the game process is determined according to the character information, and the game time delay is sent to the PC terminal. By determining the character information of the region of interest of the game time delay screenshot, the range of character information identification is reduced, the efficiency of character information identification is improved, the game time delay of the PC terminal is accurately determined according to the character information, the test efficiency of network time delay is improved, and the evaluation of network quality is facilitated. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A hardware structure schematic diagram of an edge computing device or a game perception platform related to an embodiment of the application is shown in the figure.

[0042] Figure 2 A flowchart of a first embodiment of the network quality evaluation method of the application is shown in the figure.

[0043] Figure 3 A detailed flowchart of step S20 of a second embodiment of the network quality evaluation method of the application is shown in the figure.

[0044] Figure 4 A logic structure schematic diagram of a character recognition model of the network quality evaluation method of the application is shown in the figure.

[0045] Figure 5 A flowchart of a third embodiment of the network quality evaluation method of the application is shown in the figure.

[0046] Figure 6 A schematic diagram of average time delays corresponding to different networks of the network quality evaluation method of the application is shown in the figure.

[0047] Figure 7 A logic structure schematic diagram of the network quality evaluation device of the application is shown in the figure.

[0048] Figure 8 A logic structure schematic diagram of the network quality evaluation device of the application is shown in the figure.

[0049] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0050] It should be understood that the specific embodiments described herein are only used to explain the application, and are not used to limit the application.

[0051] The main solution of the embodiment of the application is: obtaining a game time delay screenshot of a game process sent by a PC end, determining a region of interest of the game time delay screenshot; recognizing character information of the region of interest; determining a game time delay of the game process according to the character information, and sending the game time delay to the PC end.

[0052] By determining the character information of the region of interest of the game time delay screenshot, the range of character information recognition is reduced, the efficiency of character information recognition is improved, the game time delay is determined according to the character information, the game time delay of the PC end is accurately determined, the test efficiency of network time delay is improved, and the evaluation of network quality is facilitated.

[0053] As an implementation scheme, the edge computing device or the game perception platform can be as shown in Figure 1 .

[0054] The embodiment scheme of the application relates to an edge computing device or a game perception platform, and the edge computing device or the game perception platform comprises a processor 101, such as a CPU, a memory 102, and a communication bus 103. The communication bus 103 is used to realize the connection and communication between the components.

[0055] The memory 102 can be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. As shown in Figure 1 , the memory 102 of the edge computing device as a computer readable storage medium can include a network quality evaluation program; and the processor 101 can be used to call the network quality evaluation program stored in the memory 102 and perform the following operations:

[0056] obtaining a game time delay screenshot of a game process of a PC end, determining a region of interest of the game time delay screenshot;

[0057] recognizing character information of the region of interest;

[0058] determining a game time delay of the game process according to the character information, and sending the game time delay to the PC end.

[0059] In an embodiment, the processor 101 can be used to call the network quality evaluation program stored in the memory 102 and perform the following operations:

[0060] determining a text box in the region of interest and text box information corresponding to the text box, the text box information at least including coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box;

[0061] determining the text box with the first confidence degree greater than a preset first threshold value as a target text box;

[0062] The target text box is input into a preset text recognition model to obtain the character information corresponding to the target text box and the second confidence level corresponding to the character information.

[0063] In one embodiment, the processor 101 can be used to invoke a network quality evaluation program stored in the memory 102 and perform the following operations:

[0064] Character information whose second confidence level is greater than a preset second threshold is identified as target character information;

[0065] If the target character information includes a preset character, then obtain the first coordinate information of the text box corresponding to the target character information;

[0066] Obtain the second coordinate information of the other text boxes in the region of interest;

[0067] The game latency is determined based on the first coordinate information and the second coordinate information.

[0068] In one embodiment, the processor 101 can be used to invoke a network quality evaluation program stored in the memory 102 and perform the following operations:

[0069] Determine the absolute value of the distance between the first coordinate information and each of the second coordinate information;

[0070] The game latency is determined by the character information corresponding to the text box containing the minimum absolute value.

[0071] Or, such as Figure 1 As shown, the memory 102 of the game-aware platform, which serves as a computer-readable storage medium, may include a network quality evaluation program; and the processor 101 may be used to call the network quality evaluation program stored in the memory 102 and perform the following operations:

[0072] Obtain the game latency of the game process on the PC for a preset time period, and determine the average latency of the game latency during the preset time period;

[0073] Obtain network quality test data from the PC, including network jitter and packet loss.

[0074] Network quality is determined based on the average latency and the dial-up test data.

[0075] In one embodiment, the processor 101 can be used to invoke a network quality evaluation program stored in the memory 102 and perform the following operations:

[0076] Determine the network type corresponding to the network, and obtain the corresponding quality difference conditions based on the network type;

[0077] When the average time delay and the ping data satisfy the quality condition, it is determined that the network quality is poor;

[0078] When the average time delay and the ping data do not satisfy the quality condition, it is determined that the network quality is normal.

[0079] Based on the hardware architecture of the edge computing device or the game perception platform, an embodiment of the network quality evaluation method is provided.

[0080] Referring to Figure 2 , Figure 2 For the first embodiment of the network quality evaluation method, the network quality evaluation method comprises the following steps:

[0081] In step S10, a game time delay screenshot of a game process on a PC is obtained, and a region of interest of the game time delay screenshot is determined.

[0082] Specifically, the edge computing device is a CFDN (Content & Function Delivery Network), which is a cluster of network elements with edge computing capabilities upgraded from a CDN (Content Delivery Network). The CFDN is an edge computing network with content and computing capabilities for distribution and scheduling.

[0083] The edge computing device is in communication connection with a PC (personal computer) client. The PC end collects game information in a game process, wherein the game information includes game time delay screenshots and network performance indicators and the like. Optionally, the collection of game information on the PC end corresponds to a logical layer which is divided into five layers. One layer is a network card data scanning layer, which is provided with a process frame listening module for obtaining all processes on the local computer, an IP (Internet Protocol) and port grabbing module for querying the IP address and port of the game process TCP (Transmission Control Protocol) | UDP (User Datagram Protocol) in the network card in real time, and a route tracking module for obtaining a router interface list in a data packet path. One layer is a software and hardware environment layer, which is provided with a PC hardware detection module for obtaining computer system information, and an IP comprehensive service module for obtaining parameters such as the home city of a public network IP address, the receiving rate of wifi, and local area network quality monitoring interfaces. One layer is a game time delay image layer, which is provided with a game screenshot module for obtaining a current program handle, judging whether it is a game process, if so, using a python library pyqt5 to take a screenshot, converting the user screen resolution, taking a game time delay screenshot according to feature image coordinates, and finally uploading the game time delay screenshot. One layer is a time delay and packet loss detection layer, which is provided with a local time delay module netstat (network statistics) command for obtaining current computer network connections, routing tables and network interface information, pinging a local gateway to obtain local time delay and packet loss, and a provincial network time delay and OLT (optical line terminal) time delay module tracert (trace route) command for obtaining a current OLT address of a client, and then obtaining the delay and packet loss to the OLT server and the provincial network server. Another layer is a Socket real-time communication layer, which is provided with a client c++ and a python end. The client c++ performs Socket communication through an agreed port number, sends a request to the python end, the python end responds to interface data such as computer basic information, a game list and game delay according to a keyword, and the Socket communication return value is binary, which is converted into a utf-8 encoded string by the c++ end.

[0084] From the game information collection of the PC terminal, the PC terminal collects the game delay screenshot of the game process, the edge computing device obtains the game delay screenshot of the game process of the PC terminal, and the region of interest of the game delay screenshot is determined. For example, the width and height of the game delay screenshot are HxW, the coordinates of the region of interest are (x left ,y top ,width,height), wherein (x left ,y top ) represents the upper left corner coordinates of the region of interest, (width,height) represents the width and height of the region of interest, the region of interest is extracted using the extraction algorithm of the region of interest in OpenCV, and the coordinates of the extracted region of interest are

[0085] Step S20, recognizing the character information of the region of interest;

[0086] Specifically, the character information of the region of interest is recognized, wherein the character information includes text, numbers, English letters, etc. The character information of the region of interest can be recognized first, and then the character information in the text box is recognized through a preset recognition model; or the character information in the region of interest can be directly recognized through a preset recognition model.

[0087] Step S30, determining the game delay of the game process according to the character information, and sending the game delay to the PC terminal.

[0088] Specifically, the game delay of the game process is determined according to the character information, the recognized character information is traversed, and if the character information is 37ms, the game delay is determined to be 37ms. After determining the game delay corresponding to the game delay screenshot, the game delay is sent to the PC terminal, so that the game perception platform can obtain the game delay corresponding to the game process of the PC terminal.

[0089] In the technical scheme of the embodiment, the game delay screenshot of the game process sent by the PC terminal is obtained, the region of interest of the game delay screenshot is determined, the character information of the region of interest is recognized, the game delay of the game process is determined according to the character information, and the game delay is sent to the PC terminal. By determining the character information of the region of interest of the game delay screenshot, the range of character information recognition is reduced, the efficiency of character information recognition is improved, the game delay of the PC terminal is accurately determined according to the character information, the test efficiency of network delay is improved, and the evaluation of network quality is facilitated.

[0090] Referring to Figure 3 , Figure 3 The second embodiment of the network quality evaluation method of the application is based on the first embodiment, and the step S20 comprises:

[0091] Step S21, determine the text box in the region of interest and the text box information corresponding to the text box, the text box information at least includes the coordinate information, width and height information of the text box and the first confidence corresponding to the text box;

[0092] Step S22, determine the text box with the first confidence greater than the preset first threshold as the target text box;

[0093] Step S23, input the target text box into the preset text recognition model to obtain the character information corresponding to the target text box and the second confidence corresponding to the character information.

[0094] Specifically, the text box of the region of interest and the text box information corresponding to the text box are determined according to a preset algorithm, wherein the preset algorithm can be CRAFT (Character Region Awareness for Text Detection, Character Region Awareness for Text Detection) algorithm, which is used for text detection of the region of interest to obtain the text box and the text box information, wherein the text box information includes the first confidence corresponding to the text box, and the coordinate information and the width and height information bbox=(x1,y1,w,h) of the text box in the region of interest, wherein (x1,y1) represents the coordinate information of the text box, and (w,h) represents the width and height information.

[0095] Determine the text box with the first confidence greater than the preset first threshold as the target text box, input the target text box into the preset text recognition model to obtain the character information corresponding to the target text box and the second confidence corresponding to the character information. Wherein, as shown in Figure 4 The text recognition model includes a convolution layer, a cycle layer and a transcription layer, the convolution layer is used to extract the feature sequence of the target text box; the cycle layer is used to determine the label sequence corresponding to the feature sequence; the transcription layer outputs the character information and the second confidence of the target text box according to the label sequence.

[0096] Optionally, the convolution layer uses a convolutional neural network, and the convolution layer is used to extract the text features of the text box. The input text box bbox∈R h×w×3 That is, an RGB image with a height and width of h x w, a ResNet50 deep residual network is selected as a backbone network, the Resnet50 network has been pre-trained on an ImageNet dataset, and the spatial dimension is down-sampled to obtain a Conv4 layer feature map F t ∈R h′×w′×c , wherein the Conv4 layer feature map size is c is the channel dimension, c=2048.

[0097] Optionally, a deep bidirectional recurrent neural network can be built above the convolutional layer as a recurrent layer. The feature map output from the convolutional layer is used as the input to the recurrent layer. First, a 1×1 convolution is used to reduce the channel dimension of the feature map to 512, and the width and height matrix of the feature map is transformed into a one-dimensional vector, forming a feature map of size x = c×h′×w′, which is then input into the recurrent neural network. The recurrent layer provides each X = x1, x2, ..., x... T Feature sequence frame x t Predict a label y t Where t = 1, ..., T. N bidirectional LSTM (Long Short-Term Memory) networks are stacked to form a recurrent layer. Each bidirectional LSTM contains hidden layers h = (h1, h2, ..., h...). T If ), then the output will be:

[0098]

[0099]

[0100]

[0101] in, Represents the forward hidden vector. Represents the reverse hidden vector; h 0 =x, Let W represent the hidden vector at time t in the nth bidirectional LSTM, and let W represent the weight matrix, as shown below. Represents the hidden layer vector h n-1 and h n The weight matrix between them Here, H represents the bias vector, and H represents the hidden layer function. The H function is implemented using the standard LSTM. This represents the weight matrix between the Nth hidden layer and the output. Let b represent the hidden vector at time t in N bidirectional LSTMs. y This represents the output bias vector, N=2, T=256.

[0102] Optionally, the transcription layer outputs the predicted tag sequence y = y1,...,y from each frame of the recurrent layer. T As input, where T represents the sequence length, the output is the character information of the target text box and the second confidence score. Let... Let L represent the probability distribution on set L1, where L1 = L ∪ {blank}, L contains all labels from the task, and blank represents the empty character. (Based on sequence) Define a sequence-to-sequence mapping function B, where B maps π to the label sequence l, and the conditional probability is defined as the sum of all probabilities of π that can be mapped to l by B:

[0103]

[0104] wherein the probability of π is denotes the probability of label π at time stamp t. t

[0105] The training data set corresponding to the character recognition model is χ = {I i , l i} i , wherein I i is a training image, l i is a sequence label true value, i = 1,...,M, and M represents the number of samples in the training data set. The loss function of the entire network can be defined as:

[0106]

[0107] y i is the label sequence output by the convolutional layer and the recurrent layer after I i passes through the network. Based on the loss function, the network is back propagated, and the network parameters are continuously updated by the batch gradient descent method. After 100,000 times of batch training, the model converges.

[0108] According to the character information, the game delay of the game process can be determined. The character information with the second confidence greater than the preset second threshold value can be determined as target character information. If the target character information includes a preset character, the preset character can be ms. Then, the first coordinate information (x ms ,y ms ) of the text box corresponding to the target character information is obtained, wherein the first coordinate information can be the top-left corner coordinate of the text box. In addition, the second coordinate information (x k ,y k ) of other text boxes in the region of interest is obtained, wherein the second coordinate information can be the top-left corner coordinate of the other text boxes. According to the first coordinate information and the second coordinate information, the game delay is determined. Optionally, the absolute value of the distance between the first coordinate information and each second coordinate information is determined, as shown in the following formula:

[0109] (Δx k ,Δy k ) = (|x k -x ms |, |y k -y ms |);

[0110] wherein k = 1,...,K and k ≠ k ms , k ms ​The target character frame information represents a subscript value corresponding to a preset character ms, and K represents the number of output characters. k ,Δy k ) corresponds to the character information of the text frame. Optionally, the distance between the first coordinate information and each second coordinate information can be Euclidean distance and Manhattan distance, etc.

[0111] In the technical scheme of the embodiment, the text frame in the region of interest and the text frame information corresponding to the text frame are determined, the text frame information at least includes coordinate information, width and height information of the text frame, and a first confidence degree corresponding to the text frame; the text frame with the first confidence degree greater than a preset first threshold value is determined as a target text frame; the target text frame is input into a preset text recognition model to obtain character information corresponding to the target text frame and a second confidence degree corresponding to the character information. The character information corresponding to the text frame is recognized by the text recognition model, so as to determine the game delay according to the character information subsequently, accurately determine the game delay of the PC end, and facilitate the evaluation of the network quality.

[0112] Referring to Figure 5 , Figure 5 The method comprises the following steps:

[0113] In step S40, the game delay of the game process of the PC end in a preset time period is obtained, and an average delay of the game delay in the preset time period is determined.

[0114] In step S50, the network quality of the PC end is obtained by using the test data, and the test data includes network jitter and packet loss.

[0115] In step S60, the network quality is determined according to the average delay and the test data.

[0116] Specifically, the game perception platform is in communication connection with the PC, the game perception platform obtains the game delay of the game process of the PC end in a preset time period, determines an average value of the game delay in the preset time period as an average delay, and determines the network quality according to the average delay and the test data of the network quality of the PC end, thereby accurately determining the game quality problem.

[0117] The network quality is determined according to the average delay and the test data, and since different networks correspond to different average delays, for example, Figure 6As shown in the figure, curve a represents the average delay of the server, curve b represents the local average delay, curve c represents the average delay of the metropolitan area network, curve d represents the average delay of the provincial network, and the quality conditions corresponding to different networks are different, so the network type corresponding to the network is determined first, and the corresponding quality condition is obtained according to the network type; when the average delay and the dialing test data meet the quality difference condition, it is determined that the network quality is poor; when the average delay and the dialing test data do not meet the quality difference condition, it is determined that the network quality is normal.

[0118] Optionally, according to the average delay of the test PC to the access router / optical modem and the dialing test data of the network quality, it is judged whether the user local area network is abnormal, and the quality difference condition is that the average delay is greater than or equal to 5ms, the jitter is greater than 2ms, and the packet loss is 0%.

[0119] Optionally, according to the average delay of the test PC to the city metropolitan area network and the dialing test data of the network quality, it is judged whether the city metropolitan area network is poor, and the quality difference condition is that the average delay is greater than or equal to 10ms, the jitter is greater than 2ms, and the packet loss is 0%; if the quality difference appears, it is prompted to contact the city metropolitan area network to check the line or abnormal configuration.

[0120] Optionally, according to the average delay of the test PC to the provincial network equipment and the dialing test data of the network quality, it is judged whether the provincial network is poor, and the quality difference condition is that the average delay is greater than or equal to 15ms, the jitter is greater than 5ms, and the packet loss is 0%; if the quality difference appears, it is prompted to contact the provincial network administrator to check the provincial network equipment / route and the like.

[0121] Optionally, according to the average delay of the test PC to the game server and the dialing test data of the network quality, it is judged whether there is content quality difference, and the quality difference condition is that the jitter is greater than 10ms and the packet loss is 0%, wherein the average delay is judged according to the region in the following table:

[0122] Zhejiang time delay Shanghai time delay Beijing time delay Guangdong time delay Chongqing time delay Greater than 20 ms Greater than 25 ms Greater than 35 ms Greater than 40 ms Greater than 50 ms

[0123] If the quality difference appears, the PC client submits the optimization demand to the game perception platform, the game perception platform generates and issues the content quality difference scheduling strategy according to the demand, compares the quality before and after scheduling after scheduling is completed, and completes the quality difference optimization.

[0124] In the technical scheme of the embodiment, the game delay of the game process of the PC end in a preset time period is obtained; the average delay of the game delay in the preset time period is determined; the dialing test data of the network quality of the PC end is obtained, and the dialing test data includes network jitter and packet loss; and the network quality is determined according to the average delay and the dialing test data. The network quality is determined by the average delay and the dialing test data, which improves the accuracy and comprehensiveness of the network quality evaluation.

[0125] Reference Figure 7 , Figure 7The application discloses a network quality evaluation device.

[0126] A first acquisition module 100 is configured to acquire a game time delay screenshot of a game process of a PC end, and determine a region of interest of the game time delay screenshot.

[0127] An identification module 200 is configured to identify character information of the region of interest.

[0128] A determination module 300 is configured to determine a game time delay of the game process according to the character information, and send the game time delay to the PC end.

[0129] In an embodiment, in the process of identifying the character information of the region of interest, the identification module 200 is specifically configured to:

[0130] determine a text box in the region of interest and text box information corresponding to the text box, wherein the text box information at least includes coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box.

[0131] determine the text box with the first confidence degree greater than a preset first threshold value as a target text box.

[0132] input the target text box into a preset text recognition model, to obtain the character information corresponding to the target text box and a second confidence degree corresponding to the character information.

[0133] In an embodiment, in the process of determining the game time delay of the game process according to the character information, the determination module 300 is specifically configured to:

[0134] determine character information with the second confidence degree greater than a preset second threshold value as target character information.

[0135] if the target character information includes a preset character, acquire first coordinate information of a text box corresponding to the target character information.

[0136] acquire second coordinate information of other text boxes in the region of interest.

[0137] determine the game time delay according to the first coordinate information and the second coordinate information.

[0138] In an embodiment, in the process of determining the game time delay according to the first coordinate information and the second coordinate information, the determination module 300 is specifically configured to:

[0139] determine absolute values of distances of the first coordinate information and each second coordinate information.

[0140] The character information corresponding to the text box corresponding to the minimum value in the absolute values is the game delay.

[0141] Referring to Figure 8 , Figure 8 The network quality evaluation device comprises a network quality evaluation device.

[0142] The computing module 400 is configured to obtain a game delay of a game process of a PC terminal in a preset time period, and determine an average delay of the game delay in the preset time period.

[0143] The second obtaining module 500 is configured to obtain network quality test data of the PC terminal, wherein the network quality test data comprises network jitter and packet loss.

[0144] The evaluation module 600 is configured to determine network quality according to the average delay and the network quality test data.

[0145] In an embodiment, in terms of determining network quality according to the average delay and the network quality test data, the evaluation module 600 is specifically configured to:

[0146] Determine a network type corresponding to the network, and obtain a quality difference condition corresponding to the network type according to the network type;

[0147] When the average delay and the network quality test data satisfy the quality difference condition, it is determined that the network quality is poor.

[0148] When the average delay and the network quality test data do not satisfy the quality difference condition, it is determined that the network quality is normal.

[0149] The application further provides an edge computing device, which comprises a memory, a processor, and a network quality evaluation program stored in the memory and executable on the processor, and when the network quality evaluation program is executed by the processor, each step of the network quality evaluation method is implemented.

[0150] The application further provides a computer readable storage medium, which stores a network quality evaluation program, and when the network quality evaluation program is executed by a processor, each step of the network quality evaluation method is implemented.

[0151] The above-mentioned application embodiment serial numbers are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0152] It should be noted that, in the present document, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, system, article, or apparatus that comprises a list of elements does not include only those elements recited, but can also include other elements not expressly listed or inherent to such process, system, article, or apparatus. Without more limitations, an element defined by the phrase "comprising a" does not exclude the existence of additional identical elements in the process, system, article, or apparatus that includes the element.

[0153] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment system can be realized by means of software and the necessary general hardware platform, of course, it can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a computer readable storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for making a terminal device (which can be a mobile phone, computer, parking management device, air conditioner, or network device, etc.) execute the system described in each embodiment of the present application.

[0154] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent flow transformation made by using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method of evaluating network quality, characterized by, The network quality evaluation method applied to an edge computing device comprises: Obtaining a game time delay screenshot of a game process of a PC end, determining a region of interest of the game time delay screenshot; Determining a text box in the region of interest and text box information corresponding to the text box, the text box information at least including coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box; Determining the text box with the first confidence degree greater than a preset first threshold as a target text box; Inputting the target text box into a preset text recognition model to obtain character information corresponding to the target text box and a second confidence degree corresponding to the character information; Determining the character information with the second confidence degree greater than a preset second threshold as target character information; If the target character information includes a preset character, obtaining first coordinate information of the text box corresponding to the target character information; Obtaining second coordinate information of other text boxes in the region of interest; Determining a game time delay according to the first coordinate information and the second coordinate information, and sending the game time delay to the PC end.

2. The method of evaluating network quality according to claim 1, wherein The text recognition model comprises a convolution layer, a recurrent layer and a transcription layer, the convolution layer being used to extract a feature sequence of the target text box; The recurrent layer is used to determine a label sequence corresponding to the feature sequence; and the transcription layer is used to output character information of the target text box and the second confidence degree according to the label sequence.

3. The method of evaluating network quality according to claim 1, wherein, The step of determining the game time delay according to the first coordinate information and the second coordinate information comprises: Determining absolute values of distances of the first coordinate information and each second coordinate information; According to the character information corresponding to the text box corresponding to the minimum value in the absolute values, the game time delay is determined.

4. A method of evaluating network quality, characterized by, The network quality evaluation method applied to a game perception platform comprises: Obtaining a game time delay of a game process of a PC end in a preset time period, and determining an average time delay of the game time delay in the preset time period; wherein an edge computing device obtains a game time delay screenshot of a game process of a PC end, determines a region of interest of the game time delay screenshot, determines a text box in the region of interest and text box information corresponding to the text box, the text box information at least including coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box, determines the text box with the first confidence degree greater than a preset first threshold as a target text box, inputs the target text box into a preset text recognition model to obtain character information corresponding to the target text box and a second confidence degree corresponding to the character information, determines the character information with the second confidence degree greater than a preset second threshold as target character information, if the target character information includes a preset character, obtains first coordinate information of the text box corresponding to the target character information, obtains second coordinate information of other text boxes in the region of interest, and determines a game time delay according to the first coordinate information and the second coordinate information, and sends the game time delay to the PC end; Obtaining network quality test data of the PC end, the test data including network jitter and packet loss; Determine network quality according to the average latency and the ping data.

5. The method of evaluating network quality according to claim 4, wherein, The step of determining network quality according to the average latency and the ping data comprises: Determine the network type corresponding to the network, and obtain the corresponding quality condition according to the network type; When the average latency and the ping data meet the quality condition, it is determined that the network quality is poor; When the average latency and the ping data do not meet the quality condition, it is determined that the network quality is normal.

6. An apparatus for evaluating network quality, characterized by comprising: The network quality evaluation device comprises: A first acquisition module is configured to acquire a game latency screenshot of a game process of a PC terminal and determine a region of interest of the game latency screenshot. An identification module is configured to determine a text box in the region of interest and text box information corresponding to the text box, the text box information comprising at least coordinate information, width and height information of the text box, and a first confidence degree corresponding to the text box, determine the text box with the first confidence degree greater than a preset first threshold as a target text box, and input the target text box into a preset text recognition model to obtain character information corresponding to the target text box and a second confidence degree corresponding to the character information. A determination module is configured to determine character information with the second confidence degree greater than a preset second threshold as target character information, acquire first coordinate information of the text box corresponding to the target character information if the target character information comprises a preset character, acquire second coordinate information of other text boxes in the region of interest, determine a game latency according to the first coordinate information and the second coordinate information, and send the game latency to the PC terminal.

7. An edge computing device, characterized by The edge computing device comprises a memory, a processor, and a network quality evaluation program stored in the memory and executable on the processor, and the network quality evaluation program, when executed by the processor, implements each step of the network quality evaluation method according to any one of claims 1-3.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a network quality evaluation program, and the network quality evaluation program, when executed by the processor, implements each step of the network quality evaluation method according to any one of claims 1-3 or 4-5.

Images