Data processing method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a data processing method and device, computer equipment, a computer readable storage medium and a computer program product. The method is applied to a set top box, and comprises the following steps: acquiring a real-time video stream at a current moment, and performing identification processing on the real-time video stream to obtain an initial identification result; calculating a result compensation coefficient at the current moment according to a first historical recognition result corresponding to the set top box and a second historical recognition result fed back by the collaborative terminal; and performing result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and performing output display based on the target recognition result. By adopting the method, the accuracy of real-time identification of the user action can be improved.

Description

Technical Field

[0001] The present application relates to the field of body recognition technology, and in particular to a data processing method, device, computer equipment, computer-readable storage medium and computer program product. Background Art

[0002] With the development of artificial intelligence technology, IPTV (Internet Protocol TV) set-top boxes can also provide AI (Artificial Intelligence) fitness services. Through the AI ​​image analysis of the set-top box, the user's skeleton and movement recognition can be realized.

[0003] In traditional technology, if the AI ​​image analysis capability of a set-top box is limited, the set-top box can transmit the video stream to be identified to a collaborative terminal (for example, a mobile phone, etc.), and the collaborative terminal will process the video stream to be identified to obtain the identification result, and feed the identification result back to the set-top box, which will then display the identification result on the TV screen.

[0004] However, in traditional technologies, since it takes a certain amount of time for the cooperative terminal to receive the video stream and decompress and recognize the received video stream, the delay after the recognition result is fed back to the set-top box increases, which in turn leads to poor accuracy in real-time recognition of user actions. Summary of the invention

[0005] Based on this, it is necessary to provide a data processing method, apparatus, computer device, computer-readable storage medium and computer program product to address the above technical issues.

[0006] In a first aspect, the present application provides a data processing method, which is applied to a set-top box, comprising:

[0007] Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0008] Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0009] The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

[0010] In one embodiment, before obtaining the real-time video stream at the current moment and performing recognition processing on the real-time video stream to obtain an initial recognition result, the method further includes:

[0011] Acquire first performance information fed back by the cooperative terminal, and determine a target recognition model corresponding to the cooperative terminal according to the first performance information;

[0012] The model information of the target recognition model is synchronized to the cooperative terminal, and the cooperative terminal is instructed to perform recognition processing on the real-time video stream according to the target recognition model.

[0013] In one embodiment, the calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal includes:

[0014] Obtaining a first historical recognition result of the set-top box for a target action and a second historical recognition result of the cooperative terminal for the target action at a historical moment; the target action is an action to be recognized contained in the real-time video stream at the current moment;

[0015] Determine a result compensation coefficient at a current moment according to a difference between the first historical recognition result and the second historical recognition result.

[0016] In one embodiment, determining the result compensation coefficient at the current moment according to the difference between the first historical recognition result and the second historical recognition result includes:

[0017] Determining a confidence level corresponding to the collaborative terminal according to the first performance information of the collaborative terminal and the second performance information of the set-top box;

[0018] Determine a dynamic frequency adjustment factor corresponding to the historical moment according to a time difference between the historical moment and the current moment and an action time interval of the target action;

[0019] The difference between the first historical recognition result and the second historical recognition result is corrected based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at a current moment.

[0020] In one embodiment, after acquiring the real-time video stream at the current moment and performing recognition processing on the real-time video stream to obtain an initial recognition result, the method further includes:

[0021] Sending the real-time video stream to the collaborative terminal, instructing the collaborative terminal to perform recognition processing on the real-time video stream;

[0022] Receive the collaborative recognition result for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition result and the initial recognition result into a recognition result sequence; the initial recognition result serves as the first historical recognition result at the next moment, and the collaborative recognition result serves as the second historical recognition result at the next moment.

[0023] In one embodiment, performing result compensation processing on the initial recognition result based on the result compensation coefficient to obtain the target recognition result includes:

[0024] If the result compensation coefficient is a positive correction, the result compensation coefficient is added to the initial recognition result to obtain the target recognition result.

[0025] In a second aspect, the present application further provides a data processing device, which is applied to a set-top box, comprising:

[0026] A first acquisition module is used to acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0027] A calculation module, used for calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0028] The compensation module is used to perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and output and display based on the target recognition result.

[0029] In one embodiment, the device further comprises:

[0030] A second acquisition module is used to acquire first performance information fed back by the collaborative terminal, and determine a target recognition model corresponding to the collaborative terminal according to the first performance information;

[0031] A synchronization module is used to synchronize the model information of the target recognition model to the collaborative terminal, and instruct the collaborative terminal to perform recognition processing on the real-time video stream according to the target recognition model.

[0032] In one embodiment, the calculation module is specifically used to obtain a first historical recognition result of the set-top box for a target action and a second historical recognition result of the cooperative terminal for the target action at a historical moment; the target action is an action to be recognized contained in the real-time video stream at the current moment;

[0033] Determine a result compensation coefficient at a current moment according to a difference between the first historical recognition result and the second historical recognition result.

[0034] In one of the embodiments, the calculation module is specifically used to determine the confidence level corresponding to the collaborative terminal according to the first performance information of the collaborative terminal and the second performance information of the set-top box;

[0035] Determine a dynamic frequency adjustment factor corresponding to the historical moment according to a time difference between the historical moment and the current moment and an action time interval of the target action;

[0036] The difference between the first historical recognition result and the second historical recognition result is corrected based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at a current moment.

[0037] In one embodiment, the device further comprises:

[0038] A sending module, used to send the real-time video stream to the collaborative terminal, and instruct the collaborative terminal to identify and process the real-time video stream;

[0039] A receiving module is used to receive the collaborative recognition result for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition result and the initial recognition result into a recognition result sequence; the initial recognition result serves as the first historical recognition result at the next moment, and the collaborative recognition result serves as the second historical recognition result at the next moment.

[0040] In one of the embodiments, the compensation module is specifically configured to, if the result compensation coefficient is a positive correction, add the result compensation coefficient to the initial recognition result to obtain a target recognition result.

[0041] In a third aspect, the present application also provides a data processing system, comprising:

[0042] A set-top box is used to obtain a real-time video stream at a current moment, perform recognition processing on the real-time video stream to obtain an initial recognition result; calculate a result compensation coefficient at the current moment according to a first historical recognition result corresponding to the set-top box and a second historical recognition result fed back by a collaborative terminal; perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and output and display based on the target recognition result;

[0043] The collaborative terminal is used to identify and process the historical video stream of the historical moment, obtain a second historical identification result, and feed back the second historical identification result to the set-top box.

[0044] In a fourth aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0045] Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0046] Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0047] The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

[0048] In a fifth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0049] Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0050] Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0051] The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

[0052] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0053] Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0054] Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0055] The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

[0056] The above-mentioned data processing method, apparatus, computer equipment, computer-readable storage medium and computer program product calculate the result compensation coefficient for the set-top box at the current moment by using the first historical recognition result and the second historical recognition result respectively completed by the set-top box and the cooperative terminal as the basis for calculating the result compensation coefficient, so as to calibrate the initial recognition result of the set-top box at the current moment. The result compensation coefficient is not the recognition result of the cooperative terminal at the current moment, but a correction amount obtained based on the difference between the first historical recognition result and the second historical recognition result. Therefore, the correction amount is not affected by factors such as video stream packet loss and recognition delay, and can guide the correction of the initial recognition result of the set-top box, thereby improving the accuracy of real-time recognition of user actions. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 An application environment diagram of a data processing method in an embodiment;

[0059] Figure 2 is a flow chart of a data processing method in one embodiment;

[0060] Figure 3 A schematic diagram of a process of calling a collaborative capability of a set-top box in one embodiment;

[0061] Figure 4 A schematic diagram of a flow chart of calculating a result compensation coefficient in one embodiment;

[0062] Figure 5 A schematic diagram of a flow chart of introducing confidence and dynamic frequency adjustment factors to calculate result compensation coefficients in one embodiment;

[0063] Figure 6 A schematic diagram of a flow chart of interaction between a set-top box and a cooperative terminal during a cooperative identification process in one embodiment;

[0064] Figure 7 A schematic flow chart of an example of a data processing method in an embodiment;

[0065] Figure 8 is a structural block diagram of a data processing device in one embodiment;

[0066] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0068] The data processing method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. The set-top box 102 communicates with the cooperative terminal 104 through the network, and the set-top box communicates with the camera device (CAM, Camera) 106 through the network (for example, WIFI, etc.) or a wired connection (for example, USB, etc.). The set-top box 102 obtains the real-time video stream at the current moment, performs recognition processing on the real-time video stream, and obtains an initial recognition result; calculates the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal; performs result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and outputs and displays based on the target recognition result. Among them, the set-top box 102 can be an IPTV terminal (set-top box), the set-top box 102 has an AI chip for motion recognition, the set-top box 102 has built-in AI fitness application, AI processing module, video processing module and multi-terminal collaboration module, the AI ​​fitness application is used to initiate AI fitness business, and the AI ​​fitness business is an interactive fitness service realized by the IPTV set-top box and the connected camera device; the AI ​​processing module is used to identify and process the real-time video stream sent by the camera device 106, and the AI ​​processing module uses the AI ​​chip in the set-top box to perform efficient image processing and recognition, and performs positive adjustment in combination with the result compensation coefficient; the video processing module is used to compress and decompress the real-time video stream sent by the camera device 106, return the video frames of the real-time video stream according to the number of frames that the AI ​​processing module can process, and transmit the copied real-time video stream to the collaboration terminal; the multi-terminal collaboration module is used for terminal discovery, AI compensation and data transmission. The collaboration terminal 104 may include but is not limited to a personal mobile terminal or NAS (Network Attached Storage). The cooperative terminal 104 has a built-in multi-terminal cooperative module for connecting with the set-top box 102 , and cooperatively performing motion recognition on the real-time video stream, and feeding back the cooperative recognition result to the set-top box 102 .

[0069] In an exemplary embodiment, Figure 2 As shown, a data processing method is provided, which is applied to Figure 1 The set-top box in FIG. 1 is taken as an example to illustrate, and the method includes the following steps 202 to 206. Among them:

[0070] Step 202, obtain the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result.

[0071] In an embodiment of the present application, when the user starts the AI ​​fitness application in the set-top box, the AI ​​fitness application initiates the initialization of the AI ​​fitness service and loads related components, including an action recognition model, a video processing module, a multi-terminal collaboration module, etc. For example, the set-top box responds to the AI ​​fitness service and calls the action recognition model corresponding to the AI ​​fitness application. Different types of AI fitness services have action recognition models for the target actions involved in the AI ​​fitness service. At the same time, the set-top box performs terminal discovery through the multi-terminal collaboration module, discovers the collaborative terminal in the local area network where the set-top box is located and connects to the collaborative terminal, preparing for data transmission and collaborative processing. Among them, the AI ​​fitness service is to recognize user actions through camera video to realize AI fitness service applications such as skipping, jumping left and right, and squatting, and give positive feedback through the TV screen connected to the set-top box. For example, after the user completes a target action of skipping, jumping left and right or squatting, the TV screen gives positive feedback of counting plus one.

[0072] When the user applies the AI ​​fitness service, if the user's actions need to be identified, for example, when the user enables the counting function of the AI ​​fitness service, the set-top box calls and turns on the camera. The camera records and identifies the user's actions, obtains a real-time video stream containing the user's actions, and sends it to the set-top box synchronously via WIFI or USB connection, and then the set-top box obtains the real-time video stream at the current moment. Specifically, the video processing module of the set-top box receives the real-time video stream from the camera, and the video processing module processes the video stream and adjusts the frame rate to meet the requirements of the AI ​​processing module. For example, the AI ​​processing module may require 15 or 30 frames of images per second for real-time processing, so the video processing module of the set-top box adjusts the frame rate of the real-time video stream according to this requirement. At the same time, according to the requirements of the AI ​​processing module, the video processing module may also adjust the resolution of the real-time video stream to ensure the efficiency of the AI ​​chip of the set-top box in motion recognition.

[0073] Then, the video processing module forwards the processed real-time video stream to the AI ​​processing module for action recognition. The set-top box processes each frame of the image through the AI ​​processing module, extracts the user's skeleton information and recognizes the user's action, analyzes the user's action, and determines the probability value of the user completing the target action as the initial recognition result.

[0074] In a specific embodiment, the set-top box obtains the real-time video stream according to a preset period. The preset period can be the action time interval corresponding to the current AI fitness service, that is, different target actions have different action completion times. For example, if the target action is a squat, the preset period can be 1 second. If the target action is a complex action, the preset period can be the time to complete a target action. Therefore, the set-top box can use the time to complete an action once or multiple times as the preset period, continuously obtain the real-time video stream, and identify and process the real-time video stream of each preset period in real time.

[0075] Step 204: Calculate the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal.

[0076] In the embodiment of the present application, due to the performance limitation of the AI ​​chip in the set-top box, the initial recognition result may have a certain error, and the cooperative terminal (for example, mobile phone, NAS) usually has a more powerful computing power and can provide more accurate recognition results. The set-top box can evaluate the recognition result of the set-top box itself through the cooperative recognition result of the cooperative terminal. Specifically, for each preset cycle, if there is a historical recognition result corresponding to the current target action in the set-top box, the set-top box evaluates the recognition error of the set-top box AI chip through the first historical recognition result of the set-top box in the previous preset cycle and the second historical recognition result of the cooperative terminal, thereby calculating a result compensation coefficient for correcting the initial recognition result at the current moment. Optionally, the set-top box can also calculate the mean square error (MSE, Mean Squared Error) or the mean absolute error (MAE, Mean Absolute Error) based on multiple first historical recognition results and second historical recognition results in the historical recognition cycle as the result compensation coefficient at the current moment. Optionally, the calculation process of the result compensation parameter can also be performed in the cooperative terminal and fed back to the set-top box after the calculation is completed.

[0077] In an optional embodiment, if there are second historical recognition results fed back by multiple collaborative terminals in the recognition result sequence of the set-top box, the set-top box can determine the recognition error between the set-top box and each collaborative terminal based on the second historical recognition results of different collaborative terminals and the first historical recognition results, and obtain multiple initial result compensation coefficients, and use the average value of the initial result compensation coefficients as the final result compensation coefficient for error correction of the initial recognition result.

[0078] Step 206, performing result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and outputting and displaying the target recognition result.

[0079] In an embodiment of the present application, the set-top box can determine the error amplitude of the initial recognition result according to the result compensation coefficient, and then apply the result compensation coefficient to the initial recognition result, correct and compensate the initial recognition result at the current moment, and finally obtain a more accurate target recognition result. The target recognition result is the probability value of the user completing the target action after correction. Then, the set-top box compares the target recognition result with the preset threshold value to finally determine whether the user has completed the target action of the current recognition cycle. If the user has completed the target action of the current recognition cycle, the count value is increased by one and displayed on the TV screen. Alternatively, if the AI ​​fitness service is a somatosensory game, etc., the current game score is calculated and displayed according to the target recognition result.

[0080] In a specific embodiment, if the AI ​​fitness service is squat counting, the preset period of action recognition is 1 second, and the first historical recognition result in the previous preset period (last second) is 0.5 (set-top box recognition result), and the second historical recognition result is 0.6 (cooperative terminal recognition result), then the set-top box can determine that for the current action, the error of the initial recognition result at the current moment may be 0.1. Then, the set-top box corrects the initial recognition result at the current moment according to the error margin of 0.1. For example, if the initial recognition result at the current moment is 0.6, and the preset threshold for judging whether to count plus one is 0.7, then the set-top box corrects the initial recognition result of 0.6 to 0.7 according to the error margin of 0.1, and compares 0.7 as the target recognition result with the preset threshold, and concludes that the current user has completed a target action, and then performs output feedback of counting plus one on the TV screen.

[0081] In the above data processing method, the result compensation coefficient for the set-top box at the current moment is calculated by taking the first historical recognition result and the second historical recognition result respectively completed by the set-top box and the cooperative terminal as the basis for calculating the result compensation coefficient, so as to calibrate the initial recognition result of the set-top box at the current moment. The result compensation coefficient is not the recognition result of the cooperative terminal at the current moment, but a correction amount obtained based on the difference between the first historical recognition result and the second historical recognition result. Therefore, the correction amount is not affected by factors such as video stream packet loss and recognition delay, and can guide the correction of the initial recognition result of the set-top box, thereby improving the accuracy of real-time recognition of user actions.

[0082] In an exemplary embodiment, before the set-top box performs AI action recognition for terminal collaboration, it is also necessary to discover and connect to the collaborative terminal, such as Figure 3 As shown, before step 202, the method further includes steps 302 to 304. Among them:

[0083] Step 302: Obtain first performance information fed back by the cooperative terminal, and determine a target recognition model corresponding to the cooperative terminal according to the first performance information.

[0084] In an embodiment of the present application, in a home or small network environment, an IPTV set-top box may need to perform some computationally intensive tasks, such as AI processing. However, due to the hardware limitations of the set-top box, these tasks may not be completed efficiently. To this end, the set-top box needs to work in collaboration with a collaborative terminal (such as NAS, smart phones, etc.) with stronger AI processing capabilities in the local area network. The set-top box discovers collaborative terminals in the local area network through a preset communication protocol, and the preset communication protocol defines a set of protocols for communication between the set-top box and potential collaborative terminals, which are used to discover and negotiate AI processing capabilities. That is, the set-top box broadcasts a discovery request using local area network technology (e.g., mDNS, SSDP, etc.) to find collaborative terminals that may provide AI processing capabilities in the local area network. When the collaborative terminal receives a discovery request from the set-top box, if it has AI processing capabilities, it will respond to the set-top box with its AI collaborative capabilities. When the collaborative terminal responds to the set-top box about its AI collaborative capability, the collaborative terminal also feeds back the first performance information of its AI chip, including performance indicators such as TOPS (Tera Operations Per Second, unit of processor computing power, trillion operations per second) and FPS (Frames Per Second). The performance indicators in the first performance information reflect the AI ​​processing capability and efficiency of the collaborative terminal.

[0085] The model library built into the set-top box contains a variety of AI models, each with different complexity and accuracy. The set-top box selects a target recognition model that matches the capabilities of the cooperative terminal. For example, if the TOPS value of the cooperative terminal is low, a simpler recognition model is selected; if the TOPS value is high, a more complex recognition model can be selected.

[0086] Step 304: synchronize the model information of the target recognition model to the cooperative terminal, and instruct the cooperative terminal to perform recognition processing on the real-time video stream according to the target recognition model.

[0087] In an embodiment of the present application, the set-top box performs model synchronization according to a preset communication protocol, transmits the target recognition model to the collaborative terminal, and instructs the collaborative terminal to perform recognition processing according to the target recognition model after subsequently receiving the real-time video stream.

[0088] In this embodiment, by calling the recognition capability of the collaborative terminal, the collaborative terminal evaluates the recognition capability of the set-top box, and then obtains the result compensation coefficient of the set-top box for the real-time video stream at the current moment, thereby realizing collaborative compensation of the set-top box's recognition capability, which can improve the accuracy of the set-top box in recognizing user actions.

[0089] In an exemplary embodiment, Figure 4As shown, step 204 includes steps 402 to 404. Among them:

[0090] Step 402: Obtain a first historical recognition result of the set-top box for the target action and a second historical recognition result of the cooperative terminal for the target action at a historical moment.

[0091] The target action is the action to be identified contained in the real-time video stream at the current moment.

[0092] In an embodiment of the present application, when the AI ​​fitness service includes multiple target actions to be identified, as the set-top box identifies the user's actions, the set-top box's recognition result sequence includes first historical recognition results and second historical recognition results corresponding to multiple actions, respectively, wherein the first historical recognition result is the set-top box's own first historical recognition result for each action, and the second historical recognition result is the collaborative terminal's historical recognition result for each action. The set-top box takes the action to be identified in the real-time video stream at the current moment as the target action, and obtains the first historical recognition result and the second historical recognition result corresponding to the target action in the historical recognition result sequence. Through the recognition results of the same action, the collaborative terminal can be used as a reference standard. The two recognize the same action under the same environment and conditions, reducing the influence of input and environmental variables. By directly calculating the difference between the recognition results of the set-top box and the collaborative terminal, the recognition error of the set-top box can be quantified.

[0093] Step 404: Determine the result compensation coefficient at the current moment according to the difference between the first historical recognition result and the second historical recognition result.

[0094] In an embodiment of the present application, when performing action recognition on the real-time video stream of the current recognition cycle, the set-top box uses the difference between the first historical recognition result and the second historical recognition result of the previous recognition cycle stored for the target action in the recognition result sequence as the result compensation coefficient at the current moment, and the result compensation coefficient is used to compensate and correct the initial recognition result.

[0095] In this embodiment, the result compensation coefficient is calculated by the difference between the first historical recognition result and the second historical recognition result to avoid displaying the real-time collaborative recognition result of the collaborative terminal. Instead, the collaborative recognition result of the collaborative terminal is used as a delay compensation coefficient to correct the initial recognition result of the set-top box, thereby avoiding the influence of packet loss, delay, etc. of the collaborative terminal on the target recognition result, ensuring that the target recognition result can respond in a timely manner, and improving the accuracy of the target recognition result through the correction compensation of the result compensation coefficient.

[0096] In an exemplary embodiment, Figure 5 As shown, step 404 includes steps 502 to 506. Among them:

[0097] Step 502: Determine the confidence level corresponding to the cooperative terminal according to the first performance information of the cooperative terminal and the second performance information of the set-top box.

[0098] In the embodiment of the present application, the set-top box uses the ratio of the first performance information of the AI ​​chip in the cooperative terminal to the second performance information of its own AI chip as the confidence of the cooperative terminal. The confidence represents the credibility of the second historical recognition result of the cooperative terminal relative to the first historical recognition result of the set-top box, and is used to adjust the correction range of the cooperative terminal to the recognition result of the set-top box.

[0099] Step 504 , determining a dynamic frequency adjustment factor corresponding to the historical moment according to the time difference between the historical moment and the current moment and the action time interval of the target action.

[0100] In an embodiment of the present application, the set-top box determines the ratio of the action time interval of the target action to the time difference between the historical moments of the first historical recognition result and the second historical recognition result and the current moment as the dynamic frequency adjustment factor corresponding to the historical moment. The dynamic frequency adjustment factor is used to further adjust the amplitude of the correction of the initial recognition result of the real-time video stream at the current moment by the second historical recognition result according to the time difference between the historical moment and the current moment and the action time interval of the target action, thereby optimizing the accuracy of the recognition result. For example, the dynamic frequency adjustment factor reflects the correlation between the second historical recognition result and the initial recognition result of the current action. The smaller the time difference, the larger the dynamic frequency adjustment factor, and the higher the weight of the historical recognition result; the larger the time difference, the smaller the dynamic frequency adjustment factor, and the lower the weight of the historical recognition result.

[0101] Step 506: Correct the difference between the first historical recognition result and the second historical recognition result based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at the current moment.

[0102] In an embodiment of the present application, there may be differences in the historical recognition results of the set-top box and the cooperative terminal, and these differences need to be corrected through confidence and dynamic frequency adjustment factors to improve the accuracy of the current recognition results. The set-top box first determines the difference between the first historical recognition result and the second historical recognition result. Specifically, the set-top box performs a difference calculation on the first historical recognition result and the second historical recognition result to obtain the difference between the first historical recognition result and the second historical recognition result. Then, the difference is corrected according to the confidence and the dynamic frequency adjustment factor to obtain the result compensation coefficient at the current moment. Among them, the application process of correcting the initial recognition result according to the result compensation coefficient is shown in the following target recognition result calculation formula:

[0103]

[0104] in, is the target recognition result, is the initial recognition result, for The first historical recognition result at the moment (previous recognition cycle), for The second historical recognition result at time , that is, Characterize the difference between the first historical recognition result and the second historical recognition result, is the action time interval of the target action, is the time difference between the historical moment and the current moment, / is the dynamic frequency adjustment factor, is the confidence corresponding to the cooperative terminal, where It is obtained by comparing the first performance information of the AI ​​chip in the collaborative terminal with the second performance information of its own AI chip.

[0105] In a specific embodiment, an IPTV terminal (set-top box) performs AI motion services and finds that the cooperative terminal is a NAS device. Under the same model in the IPTV terminal and the NAS device, 2 frames of images can be recognized per second. The IPTV terminal has weaker performance and can recognize 1 frame of image per second. Then, the AI ​​capability ratio of the cooperative terminal to the IPTV terminal Cai=2; and the action time interval T=1s, and the recognized action is squatting. Then, an example of the calculation result of the target recognition result is shown in Table 1 below.

[0106] Table 1

[0107] time 1 2 3 4 5 6 7 #timg# 0.7 0.6 0.5 0.9 0.4 0.6 0.6 #timg# 0.8 0.7 0.56 0.95 0.6 Packet loss 0.65 #timg# 0.7 0.8 0.7 1.02 0.5 1 0.8

[0108] In this embodiment, the difference between the first historical recognition result and the second historical recognition result is corrected by confidence and dynamic frequency, so that the correction amplitude of the result compensation coefficient for the current moment is dynamically adjusted according to the relationship between the recognition result at the historical moment and the recognition result at the current moment, thereby reducing the error of the result compensation coefficient, improving the accuracy of the result compensation coefficient, and thus improving the accuracy of the target recognition result.

[0109] In an exemplary embodiment, Figure 6 As shown, after step 202, the method further includes steps 602 to 604. Among them:

[0110] Step 602: Send the real-time video stream to the cooperative terminal, and instruct the cooperative terminal to perform recognition processing on the real-time video stream.

[0111] In an embodiment of the present application, after acquiring the real-time video stream, the set-top box transmits the real-time video stream to the AI ​​processing module. At the same time, the set-top box also copies the real-time video stream and transmits it to the collaborative terminal, and instructs the collaborative terminal to identify the real-time video stream.

[0112] Step 604: Receive the collaborative recognition result for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition result and the initial recognition result into a recognition result sequence.

[0113] Among them, the initial recognition result is used as the first historical recognition result at the next moment, and the collaborative recognition result is used as the second historical recognition result at the next moment.

[0114] In an embodiment of the present application, when the collaborative terminal receives the real-time video stream and uses the collaborative terminal synchronization and connection, the model information synchronized by the set-top box determines the action recognition model for identifying and processing the real-time video stream, and identifies and processes the real-time video stream according to the action recognition model, and feeds back the collaborative recognition result to the set-top box.

[0115] After the set-top box receives the real-time video stream fed back by the collaborative terminal, it uses the collaborative recognition result as the second historical recognition result of the next recognition cycle, and uses the initial recognition result at the current moment as the first historical recognition result, which is used to calculate the result compensation coefficient for the initial recognition result at the next moment. Specifically, after the set-top box performs action recognition at the next moment and obtains the initial recognition result at the next moment, the initial recognition result at the current moment is the first historical recognition result, and the collaborative recognition result fed back by the collaborative terminal for the real-time video stream at the current moment is the second historical recognition result.

[0116] In this embodiment, the real-time video stream is sent to the cooperative terminal and the set-top box for parallel processing at the same time, and the recognition result of the cooperative terminal and the initial recognition result of the set-top box are stored in the recognition result sequence at each moment, so that the dynamic update of the historical recognition result is achieved. When the recognition is performed at the next moment, the initial recognition result and the cooperative recognition result of the previous moment can be used as historical data to calculate the result compensation coefficient to optimize the recognition result at the current moment, which not only avoids the calculation delay of the target recognition result through parallel processing, but also improves the accuracy of the action recognition of the set-top box.

[0117] In an exemplary embodiment, step 206 includes step 2061. In which:

[0118] Step 2061, if the result compensation coefficient is a positive correction, the result compensation coefficient is added to the initial recognition result to obtain the target recognition result.

[0119] In an embodiment of the present application, if the result compensation coefficient is a positive number, it indicates that the initial recognition result needs to be adjusted positively, that is, the value of the initial recognition result needs to be increased. If the result compensation coefficient is a negative number, it indicates that the initial recognition result needs to be adjusted negatively, but since the motion recognition of the AI ​​fitness service is used for counting or score calculation of somatosensory games, in order to improve the user experience, the set-top box will abandon the negative adjustment, that is, the value of the initial recognition result will not be reduced. If the result compensation coefficient is a positive correction, the result compensation coefficient is directly superimposed on the initial recognition result to obtain the optimized target recognition result, that is, the target recognition result = initial recognition result + result compensation coefficient.

[0120] In this embodiment, the target recognition result is obtained by superimposing the result compensation coefficient onto the initial recognition result. The recognition result can be dynamically adjusted according to the positive or negative condition of the result compensation coefficient, thereby obtaining a more accurate target recognition result.

[0121] In an exemplary embodiment, an example of a data processing method is provided, such as Figure 7 As shown, including:

[0122] IPTV set-top box side:

[0123] Step 711: When the IPTV set-top box starts the AI ​​fitness service, collaborative terminal discovery is performed.

[0124] Step 712: The IPTV set-top box selects a suitable model NNM according to the AI ​​processing capability of the collaborative terminal, and synchronizes it to the collaborative terminal.

[0125] Step 713: The IPTV set-top box imports the video data of the USB or WiFi camera, and copies the video data to transmit to the cooperative terminal.

[0126] Step 714: The IPTV set-top box performs image processing on the video based on its own AI capabilities, such as reducing the frame rate, image scaling, etc.

[0127] Step 715, the IPTV set-top box performs user skeleton recognition on the processed image frame, and performs action recognition based on the previous and next frames to obtain a recognition result RR (action recognition, detection rate, and completion rate).

[0128] Step 716, after the IPTV set-top box recognizes the user action, it needs to determine whether the action is known. If the action is not known, execute step 717; if the action is known (the recognition result sequence contains recognition result RR and recognition result RRc corresponding to the action), execute step 719.

[0129] Step 717: The IPTV set-top box obtains the recognition result RR recognition result RRc.

[0130] Step 718: The IPTV set-top box records the recognition result RR and the recognition result RRc into the recognition result sequence.

[0131] In step 719, the IPTV set-top box counts the differences in delteion rate and completion rate of the same action recognition in the recognition result RR and action recognition in RRc, calculates the result compensation coefficient according to the confidence and dynamic frequency adjustment factor, corrects the recognition result RR according to the result compensation coefficient, and obtains the target recognition result.

[0132] Step 720: The IPTV set-top box feeds back the target recognition result on the screen.

[0133] Collaborative terminal side:

[0134] Step 721, the collaborative terminal responds to the IPTV set-top box with its AI collaborative capability AIC (TOPS, FPS).

[0135] Step 722: The collaborative terminal receives the model synchronized with the IPTV set-top box.

[0136] Step 723: The collaborative terminal receives the video data transmitted by the IPTV set-top box.

[0137] Step 724, the cooperative terminal processes the transmitted video and performs image processing according to its own capabilities. If the processing capability is insufficient, it can only process frames other than those processed by the IPTV set-top box or perform image scaling adjustment, but the corresponding confidence of the cooperative terminal needs to be reduced.

[0138] Step 725 , the cooperative terminal performs user skeleton recognition on the remaining processed image frames, and performs recognition based on the previous and next frames to obtain a recognition result RRc (action recognition, detection rate; completion rate).

[0139] In step 726, the cooperative terminal feeds back the recognition result RRc used for AI compensation calculation to the set-top box.

[0140] In an exemplary embodiment, a data processing system is provided, the system comprising:

[0141] The set-top box is used to obtain the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result; calculate the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal; perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and output and display based on the target recognition result;

[0142] The collaborative terminal is used to identify and process the historical video stream of the historical moment, obtain a second historical identification result, and feed back the second historical identification result to the set-top box.

[0143] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0144] Based on the same inventive concept, the embodiment of the present application also provides a data processing device for implementing the data processing method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more data processing device embodiments provided below can refer to the limitations on the data processing method above, and will not be repeated here.

[0145] In an exemplary embodiment, Figure 8 As shown, a data processing device 800 is provided, comprising: a first acquisition module 801, a calculation module 802 and a compensation module 803, wherein:

[0146] The first acquisition module 801 is used to acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0147] A calculation module 802, configured to calculate a result compensation coefficient at a current moment according to a first historical recognition result corresponding to the set-top box and a second historical recognition result fed back by the cooperative terminal;

[0148] The compensation module 803 is used to perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain the target recognition result, and output and display based on the target recognition result.

[0149] In one embodiment, the apparatus 800 further includes:

[0150] A second acquisition module is used to acquire first performance information fed back by the collaborative terminal, and determine a target recognition model corresponding to the collaborative terminal according to the first performance information;

[0151] The synchronization module is used to synchronize the model information of the target recognition model to the collaborative terminal, and instruct the collaborative terminal to perform recognition processing on the real-time video stream according to the target recognition model.

[0152] In one embodiment, the calculation module 802 is specifically used to obtain a first historical recognition result of a set-top box for a target action and a second historical recognition result of a cooperative terminal for a target action at a historical moment; the target action is an action to be recognized contained in a real-time video stream at a current moment;

[0153] A result compensation coefficient at a current moment is determined according to a difference between the first historical recognition result and the second historical recognition result.

[0154] In one of the embodiments, the calculation module 802 is specifically used to determine the confidence level corresponding to the cooperative terminal according to the first performance information of the cooperative terminal and the second performance information of the set-top box;

[0155] Determine the dynamic frequency adjustment factor corresponding to the historical moment according to the time difference between the historical moment and the current moment and the action time interval of the target action;

[0156] The difference between the first historical recognition result and the second historical recognition result is corrected based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at the current moment.

[0157] In one embodiment, the apparatus 800 further includes:

[0158] A sending module, used to send the real-time video stream to the collaborative terminal, and instruct the collaborative terminal to identify and process the real-time video stream;

[0159] The receiving module is used to receive the collaborative recognition results for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition results and the initial recognition results into the recognition result sequence; the initial recognition result is used as the first historical recognition result at the next moment, and the collaborative recognition result is used as the second historical recognition result at the next moment.

[0160] In one embodiment, the compensation module 803 is specifically configured to add the result compensation coefficient to the initial recognition result to obtain the target recognition result if the result compensation coefficient is a positive correction.

[0161] Each module in the above data processing device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0162] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a data processing method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0163] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0165] Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result;

[0166] Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal;

[0167] The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

[0168] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0169] Acquire first performance information fed back by the cooperative terminal, and determine a target recognition model corresponding to the cooperative terminal according to the first performance information;

[0170] The model information of the target recognition model is synchronized to the cooperative terminal, and the cooperative terminal is instructed to perform recognition processing on the real-time video stream according to the target recognition model.

[0171] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0172] Obtaining a first historical recognition result of the set-top box for a target action and a second historical recognition result of the cooperative terminal for the target action at a historical moment; the target action is an action to be recognized contained in the real-time video stream at the current moment;

[0173] Determine a result compensation coefficient at a current moment according to a difference between the first historical recognition result and the second historical recognition result.

[0174] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0175] Determining a confidence level corresponding to the collaborative terminal according to the first performance information of the collaborative terminal and the second performance information of the set-top box;

[0176] Determine a dynamic frequency adjustment factor corresponding to the historical moment according to a time difference between the historical moment and the current moment and an action time interval of the target action;

[0177] The difference between the first historical recognition result and the second historical recognition result is corrected based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at a current moment.

[0178] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0179] Sending the real-time video stream to the collaborative terminal, instructing the collaborative terminal to perform recognition processing on the real-time video stream;

[0180] Receive the collaborative recognition result for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition result and the initial recognition result into a recognition result sequence; the initial recognition result serves as the first historical recognition result at the next moment, and the collaborative recognition result serves as the second historical recognition result at the next moment.

[0181] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0182] If the result compensation coefficient is a positive correction, the result compensation coefficient is added to the initial recognition result to obtain the target recognition result.

[0183] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0184] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0185] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0186] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0187] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0188] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A data processing method, characterized in that: The method is applied to a set-top box, and the method comprises: Acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result; Calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal; The initial recognition result is subjected to result compensation processing based on the result compensation coefficient to obtain a target recognition result, and output is performed based on the target recognition result.

2. The method according to claim 1, characterized in that Before obtaining the real-time video stream at the current moment and performing recognition processing on the real-time video stream to obtain an initial recognition result, the method further includes: Acquire first performance information fed back by the cooperative terminal, and determine a target recognition model corresponding to the cooperative terminal according to the first performance information; The model information of the target recognition model is synchronized to the cooperative terminal, and the cooperative terminal is instructed to perform recognition processing on the real-time video stream according to the target recognition model.

3. The method according to claim 1, characterized in that The calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal includes: Obtaining a first historical recognition result of the set-top box for a target action and a second historical recognition result of the cooperative terminal for the target action at a historical moment; the target action is an action to be recognized contained in the real-time video stream at the current moment; Determine a result compensation coefficient at a current moment according to a difference between the first historical recognition result and the second historical recognition result.

4. The method according to claim 3, characterized in that The determining the result compensation coefficient at the current moment according to the difference between the first historical recognition result and the second historical recognition result includes: Determining a confidence level corresponding to the collaborative terminal according to the first performance information of the collaborative terminal and the second performance information of the set-top box; Determine a dynamic frequency adjustment factor corresponding to the historical moment according to a time difference between the historical moment and the current moment and an action time interval of the target action; The difference between the first historical recognition result and the second historical recognition result is corrected based on the confidence level and the dynamic frequency adjustment factor to obtain a result compensation coefficient at a current moment.

5. The method according to claim 1, characterized in that After obtaining the real-time video stream at the current moment, performing recognition processing on the real-time video stream, and obtaining an initial recognition result, the method further includes: Sending the real-time video stream to the collaborative terminal, instructing the collaborative terminal to perform recognition processing on the real-time video stream; Receive the collaborative recognition result for the real-time video stream fed back by the collaborative terminal, and store the collaborative recognition result and the initial recognition result into a recognition result sequence; the initial recognition result serves as the first historical recognition result at the next moment, and the collaborative recognition result serves as the second historical recognition result at the next moment.

6. The method according to claim 1, characterized in that The performing result compensation processing on the initial recognition result based on the result compensation coefficient to obtain the target recognition result includes: If the result compensation coefficient is a positive correction, the result compensation coefficient is added to the initial recognition result to obtain the target recognition result.

7. A data processing system, characterized in that: The system comprises: A set-top box is used to obtain a real-time video stream at a current moment, perform recognition processing on the real-time video stream to obtain an initial recognition result; calculate a result compensation coefficient at the current moment according to a first historical recognition result corresponding to the set-top box and a second historical recognition result fed back by a collaborative terminal; perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and output and display based on the target recognition result; The collaborative terminal is used to identify and process the historical video stream of the historical moment, obtain a second historical identification result, and feed back the second historical identification result to the set-top box.

8. A data processing device, characterized in that: The device comprises: A first acquisition module is used to acquire the real-time video stream at the current moment, perform recognition processing on the real-time video stream, and obtain an initial recognition result; A calculation module, used for calculating the result compensation coefficient at the current moment according to the first historical recognition result corresponding to the set-top box and the second historical recognition result fed back by the cooperative terminal; The compensation module is used to perform result compensation processing on the initial recognition result based on the result compensation coefficient to obtain a target recognition result, and output and display based on the target recognition result.

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

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

Images