Method for processing services of a dual-recording device and related apparatus

Abstract

This application discloses a service processing method and related apparatus for a dual-recording device. The method is applied to the central application of the dual-recording device, which also includes an operator container. The operator container includes multiple categories of intelligent operators, and images of these intelligent operators are pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer. The method includes: receiving a service processing request from the host computer; acquiring target data according to the service processing task via a first communication protocol and / or a second communication protocol; determining a target intelligent operator in the operator container according to the service processing task; transmitting the target data to the target intelligent operator via the first communication protocol and / or the second communication protocol; acquiring a first result obtained by the target intelligent operator processing the target data via the first communication protocol and / or the second communication protocol; and transmitting the first result to the host computer. This application can improve the service response speed and service processing efficiency of the dual-recording device.

Description

Technical Field

[0001] This application relates to the field of electronic technology, and in particular to a business processing method and related apparatus for a dual recording device. Background Technology

[0002] In the supervision of financial transactions by financial institutions, simultaneous audio and video recording (also known as "dual recording") is required to comprehensively and objectively record key information at each stage. Traditional dual recording quality inspection methods typically involve manual review or real-time quality inspection algorithms deployed on host computers or servers. Manual review is inefficient and struggles to handle large-scale transactions. Furthermore, host computers or servers, relying on network transmission of audio and video streams, are prone to network latency, delayed responses, and excessive system resource consumption, affecting the stability of the dual recording process and resulting in inaccurate, timely, and comprehensive dual recording quality inspection. Summary of the Invention

[0003] This application provides a service processing method and related apparatus for a dual-recording device to improve the service response speed and service processing efficiency of the dual-recording device.

[0004] In a first aspect, embodiments of this application provide a service processing method for a dual-recording device, applied to the central application of the dual-recording device. The dual-recording device further includes an operator container, which includes multiple categories of intelligent operators. A single category of intelligent operator is used to process a single dual-recording service. Mirrors of the multiple categories of intelligent operators are pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer, including:

[0005] Receive a service processing request from the host computer, the service processing request including a service processing task;

[0006] According to the business processing task, target data is acquired through a first communication protocol and / or a second communication protocol; and, according to the business processing task, a target intelligent operator is determined in the operator container.

[0007] The target data is transmitted to the target intelligent operator via a first communication protocol and / or a second communication protocol.

[0008] The first result obtained by the target intelligent operator processing the target data is obtained through the first communication protocol and / or the second communication protocol;

[0009] The first result is transmitted to the host computer.

[0010] The central application includes a software tool development kit, and after obtaining the first result obtained by the target intelligent operator processing the target data, the method further includes:

[0011] The target accuracy of obtaining the first result;

[0012] If the target accuracy is detected to be less than the first preset accuracy, upgrade feedback information is determined based on the target accuracy.

[0013] The upgrade feedback information is sent to the host computer.

[0014] The software development kit is used to receive upgrade data sent by the host computer.

[0015] The target intelligent operator is updated based on the upgrade data.

[0016] The step of determining the upgrade feedback information based on the target accuracy includes:

[0017] Determine the fluctuation range of the target accuracy;

[0018] Determine the performance metrics of the target intelligent operator in processing the target data;

[0019] The upgrade range is determined based on the fluctuation range and the performance indicators;

[0020] The upgrade feedback information is generated based on the upgrade scope.

[0021] The step of performing an update operation on the target intelligent operator based on the upgrade data includes:

[0022] If the upgrade data is detected to be parameter update data for the target intelligent operator, then the parameters of the target intelligent operator are updated in the operator container according to the parameter update data.

[0023] The step of performing an update operation on the target intelligent operator based on the upgrade data includes:

[0024] If the upgrade data is detected to be the image data to be updated of the target smart operator, then the target smart operator is deleted from the operator container;

[0025] Based on the image data to be updated, the image is deployed in the operator container.

[0026] The method further includes:

[0027] If the target accuracy is detected to be less than the second preset accuracy, then at least one reference intelligent operator is determined in the operator container based on the target intelligent operator and the business processing task, and the second preset accuracy is greater than the first preset accuracy.

[0028] A business processing model is constructed based on the target intelligent operator and the at least one reference intelligent operator;

[0029] The target data is transmitted to the service processing model via a first communication protocol and / or a second communication protocol.

[0030] The second result obtained by the business processing model from processing the target data is obtained through the first communication protocol and / or the second communication protocol;

[0031] The second result is transmitted to the host computer.

[0032] The step of constructing a business processing model based on the target intelligent operator and the at least one reference intelligent operator includes:

[0033] A first algorithm framework for the target intelligent operator is determined; and at least one second algorithm framework for the at least one reference intelligent operator is determined.

[0034] Based on the first algorithm framework and the at least one second algorithm framework, the data dependency relationship between the target intelligent operator and the at least one reference intelligent operator is determined;

[0035] Acquire the computing power resources of the target intelligent operator and the at least one reference intelligent operator;

[0036] Based on the data dependencies and computing resources, the business processing model is constructed.

[0037] Secondly, embodiments of this application provide a service processing apparatus for a dual-recording device, comprising:

[0038] A central application for a dual-recording device, the dual-recording device further includes an operator container, the operator container including multiple categories of intelligent operators, a single category of intelligent operator for processing a single dual-recording service, and images of the multiple categories of intelligent operators pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer. The method includes:

[0039] A receiving unit is configured to receive a service processing request from the host computer, wherein the service processing request includes a service processing task.

[0040] The determining unit is configured to acquire target data according to the business processing task via a first communication protocol and / or a second communication protocol; and to determine a target intelligent operator in the operator container according to the business processing task.

[0041] The first transmission unit is used to transmit the target data to the target intelligent operator through a first communication protocol and / or a second communication protocol;

[0042] The acquisition unit is configured to acquire, through a first communication protocol and / or a second communication protocol, the first result obtained by the target intelligent operator processing the target data;

[0043] The second transmission unit is used to transmit the first result to the host computer.

[0044] Thirdly, embodiments of this application provide a dual-recording device, including a memory, a processor, and executable program code stored in the memory and executable on the processor. When the processor executes the executable program code, it performs the steps of the method described in the first aspect.

[0045] Fourthly, embodiments of this application provide a computer-readable storage medium storing executable program code, the executable program code including execution instructions for performing the steps of the method as described in the first aspect.

[0046] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps described in the first aspect of embodiments of this application. The computer program product may be a software installation package.

[0047] As can be seen, in this embodiment, a service processing request from the host computer is first received, the service processing request including a service processing task; then, according to the service processing task, target data is obtained through a first communication protocol and / or a second communication protocol; and, according to the service processing task, a target intelligent operator is determined in the operator container; then, the target data is transmitted to the target intelligent operator through the first communication protocol and / or the second communication protocol; then, a first result obtained by the target intelligent operator processing the target data is obtained through the first communication protocol and / or the second communication protocol; finally, the first result is transmitted to the host computer.

[0048] This application pre-deploys various intelligent operator images into dual-recording devices using operator containers. The central application transmits the acquired target data to the target intelligent operator for processing through different communication methods, and returns the processing results to the central application. This enables the dual-recording service to achieve efficient data processing and business logic execution through the collaborative architecture of operator containerization deployment, flexible communication mechanisms, and intelligent processing capabilities. Furthermore, by handling business processing requests locally, data security is improved, and processing delays caused by network fluctuations or excessive load are avoided. At the same time, the business processing flow is simplified, thereby improving business response speed and business processing efficiency. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This application provides a system architecture diagram of a business processing system.

[0051] Figure 2 This is an architectural diagram of a dual-recording device provided in an embodiment of this application;

[0052] Figure 3 This is a flowchart illustrating a service processing method for a dual-recording device provided in an embodiment of this application;

[0053] Figure 4 This is a schematic diagram of the structure of a microphone array provided in an embodiment of this application;

[0054] Figure 5 This is a schematic diagram of a sound source localization scenario provided in an embodiment of this application;

[0055] Figure 6 This is a functional unit block diagram of a service processing device for a dual-recording equipment provided in an embodiment of this application;

[0056] Figure 7 This is a functional unit block diagram of another dual-recording device's service processing apparatus provided in this application embodiment;

[0057] Figure 8 This is a schematic diagram of the structure of a dual recording device proposed in an embodiment of this application. Detailed Implementation

[0058] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0059] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0060] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0061] In the supervision of financial institutions' business transactions, simultaneous audio and video recording is required to record key information at each stage completely and objectively; this is known as "dual recording." Traditional dual recording quality inspection methods typically involve manual review or real-time quality inspection algorithms deployed on a host computer or server.

[0062] Among these issues, manual review is inefficient and struggles to handle large-scale operations. Meanwhile, the host computer or server relies on network transmission of audio and video streams, which can lead to network latency, untimely response, and excessive system resource consumption, affecting the stability of the dual recording process and resulting in inaccurate, untimely, and incomplete quality inspection of dual recording.

[0063] To address the aforementioned issues, this application provides a service processing method and related apparatus for a dual-recording device. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0064] Please see Figure 1 , Figure 1 This is a system architecture diagram of a business processing system provided in an embodiment of this application. For example... Figure 1 As shown, the business processing system 100 includes a host computer 101 and a dual recording device 102. The dual recording device 102 includes an operator container 1021, a central application 1022, a digital signal processing module 1023, and an image signal processing module 1024. The dual recording device 102 also includes peripheral hardware such as a wide-angle camera 1025, a human verification camera 1026, an LED indicator 1027, a speaker 1028, and a microphone array 1029.

[0065] Among them, the host computer 101 can be a cloud platform, server, mobile smart terminal, etc., and is responsible for initiating business processing requests, such as frame detection requests, voice playback requests, etc.

[0066] The host computer 101 communicates with the image signal processing module 1024 via the Universal Video Class (UVC) protocol and with the digital signal processing module 1023 via the USB Audio Class (UAC) protocol.

[0067] The digital signal processing module 1023 is connected to the speaker 1028 and the microphone array 1029. The microphone array 1029 is the sound input terminal and is a sound-to-electric converter used to convert sound waves in the air into analog electrical signals.

[0068] Among them, the digital signal processing module 1023 is the signal processing center, responsible for digital processing and optimization of audio signals.

[0069] The analog electrical signal output by the microphone array 1029 can first be converted into a digital signal by an analog-to-digital converter, and then transmitted to the digital signal processing module 1023. The digital signal processing module 1023 processes the digital signal through algorithms such as noise reduction, echo cancellation, volume adjustment, and equalizer adjustment, and finally outputs the processed digital signal, which is then converted back into an analog signal by a digital-to-analog converter.

[0070] Among them, the speaker 1028 is the sound output terminal and is an electroacoustic converter. It is used to receive the analog signal transmitted by the digital signal processing module 1023 and convert it into mechanical vibration, thereby driving the air to generate sound waves and restore it to audible sound.

[0071] Specifically, the digital signal processing module 1023 first processes the raw audio signal collected by the microphone array 1029 using algorithms, and then transmits the processed audio data to the host computer 101 so that the host computer 101 can perform recording and storage operations. At the same time, the host computer 101 can also send audio data to the digital signal processing module 1023. After receiving the audio data, the digital signal processing module 1023 performs amplification, equalization and other processing on the audio data according to preset parameters, and performs digital-to-analog conversion. After the digital-to-analog conversion is completed, the audio is output through the speaker 1028 to achieve playback.

[0072] The image signal processing module 1024 is connected to the wide-angle camera 1025 and the identity verification camera 1026 via the Mobile Industry Processor Interface (MIPI). The wide-angle camera 1025 and the identity verification camera 1026 are image sensor carriers. They capture light through the lens, convert the light signal into an electrical signal by the internal sensor, output the original image, and transmit the original image to the image signal processing module 1024.

[0073] The image signal processing module 1024 receives and optimizes the original image, for example, by reducing noise to reduce electronic interference, correcting color deviation by white balance, and adjusting exposure parameters to balance brightness and darkness. Finally, it outputs image data that conforms to the standard format and then transmits it to the host computer 101 through the UVC protocol so that the host computer 101 can perform operations such as image storage or video storage.

[0074] The image signal processing module 1024 performs different image optimization operations for the original images output by different cameras.

[0075] The host computer 101 communicates with the central application 1022 via a serial port protocol. The central application 1022 communicates with the operator container 1021 via the socket protocol and the real-time streaming protocol (RTSP); and the central application 1022 communicates with the image signal processing module 1024 via the RTSP protocol and with the digital signal processing module 1023 via the socket protocol.

[0076] Specifically, the host computer 101 sends a service processing request to the central application 1022 via a serial port protocol. The central application 1022 receives the service processing request from the host computer 101, determines the intelligent operator corresponding to the service processing request in the operator container 1021, and obtains the required audio and video data streams from the digital signal processing module 1023 and / or the image signal processing module 1024 via the RTSP protocol and / or Socket protocol. The required audio and video data streams are then transmitted to the intelligent operator via the Socket protocol and / or RTSP protocol. The intelligent operator performs service processing tasks on the audio and video data and returns the results to the central application 1022. The central application 1022 then sends the results to the host computer 101, or, based on the results, controls the peripheral hardware to perform corresponding operations.

[0077] The central application 1022 is connected to the LED indicator 1027. Before each startup, the dual-recording device 102 performs a functional self-test on each module and outputs the self-test results. Based on the self-test results, the central application 1022 intelligently controls the LED indicator 1027 to display different colors, allowing for quick and intuitive judgment of the operating status of each module and enabling visualized monitoring of the device's health status. For example, if the LED indicator 1027 is green, it indicates that all modules have passed the self-test and the device can start normally; if the LED indicator 1027 is red, it indicates a module malfunction that requires troubleshooting and repair.

[0078] Please refer to Figure 2 , Figure 2 This is an architectural diagram of a dual-recording device provided in an embodiment of this application. Figure 2 As shown, the dual-recording device architecture includes a hardware layer and a software layer. The hardware layer includes a system-on-chip (SOC), a camera module, a microphone array, a speaker, and LED indicators. The SOC chip runs an embedded operating system.

[0079] The software layer controls the hardware drivers; the embedded operating system allocates computing power and memory space, coordinates hardware operations within the software layer, enabling multiple tasks to run simultaneously and supporting the application layer's runtime environment; the embedded operating system provides multiple application programming interfaces (APIs) to the application layer, allowing it to implement various functions based on business needs. Specifically, for example, using a Docker container engine paired with embedded intelligent operators addresses the business needs of localized real-time quality inspection; using dual-camera applications addresses the business needs of multi-camera collaboration, while also enabling dynamic detection to detect moving objects in front of the device, and proactively controlling the column to automatically rotate and track the image based on the detection results, adding hardware awareness and enhancing proactive monitoring capabilities; using intelligent noise reduction applications addresses the business needs of signal quality optimization, achieving precise processing of audio signals and providing high-quality audio input for subsequent functions such as voice wake-up, voice recognition, and audio recording, improving the reliability of voice interaction and the accuracy of content recording; and using hardware detection applications addresses the business needs of system stability assurance, enabling status awareness, fault diagnosis, and early warning of hardware modules.

[0080] The application layer also includes a central application, which is used to schedule the operation of components such as Docker containers, dual cameras, microphone arrays, speakers and LED indicators according to business needs, so as to achieve efficient collaboration and data flow between components and ensure the smooth operation of the overall business process.

[0081] Based on this, this application provides a service processing method and related apparatus for a dual-recording device, which will be described in detail below with reference to the accompanying drawings.

[0082] Please see Figure 3 , Figure 3 This is a flowchart illustrating a service processing method for a dual-recording device provided in an embodiment of this application, as shown below. Figure 3 As shown, this method is applied to the central application of a dual-recording device. The dual-recording device also includes an operator container, which includes multiple categories of intelligent operators. A single category of intelligent operator is used to process a single dual-recording service. Mirrors of the multiple categories of intelligent operators are pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer. The method includes the following steps:

[0083] S210, receive the service processing request from the host computer, the service processing request including a service processing task.

[0084] Among them, intelligent operators are functional units that integrate artificial intelligence (AI) models or intelligent algorithms. They have built-in pre-trained models or optimization algorithms and encapsulate complex AI logic, such as deep learning model inference and feature extraction, into modular components to complete specific intelligent tasks. They can be scheduled and invoked by operator containers. For example, a face recognition intelligent operator encapsulates a face detection and feature comparison model, and can output a judgment on "whether it is the target person" from the input image.

[0085] Among them, intelligent operators reduce computational complexity while ensuring accuracy through lightweight optimization operations such as model quantization and pruning, and can be reused by multiple business scenarios at the same time.

[0086] Among them, the operator container is a standardized environment used to host, schedule, and manage intelligent operators, providing the resources required for the operation of operators and implementing lifecycle management for intelligent operators.

[0087] Specifically, the operator container can dynamically adjust resource usage based on the priority and real-time requirements of intelligent operators, ensuring stability when multiple intelligent operators run concurrently. Furthermore, the operator container isolates the execution spaces of different intelligent operators, preventing anomalies in one intelligent operator from affecting the stability of other operators or the entire system. Finally, the operator container coordinates the execution order of multiple intelligent operators and handles data interactions between operators.

[0088] In this embodiment, the operator container can be a Docker container. The Docker container encapsulates the application and its dependent environment through a container image, packaging the intelligent operator and its dependencies, such as AI frameworks, library files, configuration parameters, etc., into a Docker image to form a standardized and portable unit, ensuring that the intelligent operator can run in the same way in any environment that supports Docker.

[0089] The host computer application communicates with the central application via a serial port protocol.

[0090] Each business processing request includes a business processing task. The business processing task is the specific action requirement contained in the request, i.e., the operation that needs to be actually performed. If the business processing request is a field detection request, the business processing task is a field detection task, used to determine whether there are more than two people in the frame during the business processing; if more than two are present, an alarm message is issued. If the business processing request is a Text-to-Speech (TTS) request, the business processing task is to convert text information into speech and control the speaker to broadcast the speech. If the business processing request is an Automatic Speech Recognition (ASR) request, the business processing task is to convert speech into text information.

[0091] The business processing request includes multiple business processing tasks, such as in-frame detection and identity verification.

[0092] S220, according to the business processing task, acquire target data through a first communication protocol and / or a second communication protocol; and, according to the business processing task, determine a target intelligent operator in the operator container.

[0093] The first communication protocol refers to the RTSP protocol, and the second communication protocol refers to the Socket protocol.

[0094] The target data includes video data, and / or audio data, and / or text data.

[0095] Video data refers to real-time video data captured by wide-angle cameras and identity verification cameras, which is acquired in real-time via the RTSP protocol. Audio data refers to microphone recording data and speaker playback data, which is acquired in real-time via the Socket protocol, or playback data is sent to the speaker via the Socket protocol to play TTS audio. Text data refers to the text content sent from the host computer to the central application, which is acquired via the Socket protocol.

[0096] Different business processing tasks have their specific data requirements. For example, the in-frame detection task requires video data, the TTS task requires text data, and the ASR task requires audio data.

[0097] The AI ​​operators are pre-deployed in the dual-recording device. The specific deployment method includes the following steps:

[0098] First, store the AI ​​operator image in the directory of the device system firmware source code. Then, design the automatic deployment script for the AI ​​operator and add it to the startup items of the device system.

[0099] Then the device system firmware is compiled. At this point, the device system firmware already contains the image of the AI ​​operator, but it has not been deployed to the operator container.

[0100] Then, the system firmware is burned into the device using a firmware burning tool.

[0101] Finally, all dual-recording devices are powered on once before leaving the factory. During startup, a list of programs or scripts is automatically executed to begin deploying AI operators. The script determines whether an AI operator has been deployed based on its ID in the operator container. If not, it executes the image deployment command according to the script content; otherwise, it skips the process.

[0102] Successful deployment will change the LED indicator light from yellow to off-white. The color change of the LED indicator light after successful deployment can be customized.

[0103] Specifically, the target AI operator is determined based on the business processing task, and then the target AI operator is activated. For example, for a frame-in-frame detection task, the target AI operator is the frame-in-frame detection operator; for a TTS task, the target AI operator is the TTS operator.

[0104] S230, the target data is transmitted to the target intelligent operator through the first communication protocol and / or the second communication protocol.

[0105] If the target data includes video data, the video data is transmitted to the target intelligent operator via the RTSP protocol. If there are multiple business processing tasks, the video data is simultaneously provided to multiple AI operators via the RTSP protocol. For example, the video data is simultaneously sent to the frame detection operator and the identity verification operator via the RTSP protocol.

[0106] If the target data includes audio data, the audio data is transmitted to the target AI operator via the Socket protocol. If the target data includes both video and audio data, the video data is transmitted to the corresponding AI operator via the RTSP protocol, and the audio data is transmitted to the corresponding AI operator via the Socket protocol.

[0107] If the target data includes text data, the text data is transmitted to the target AI operator via the Socket protocol.

[0108] In one possible embodiment, in a dual-recording device, the AI ​​operator can acquire peripheral data for calculation to achieve real-time quality inspection, or it can control peripherals for intelligent operation through a central application.

[0109] For example, the wide-angle camera is located on the mechanical column of the dual-recording device. The AI ​​operator can recognize the customer's face based on the customer's photo, and then control the column to automatically rotate and track the camera through the central application, ensuring that the customer is within the range of the wide-angle camera.

[0110] S240, obtain the first result obtained by the target intelligent operator processing the target data through the first communication protocol and / or the second communication protocol.

[0111] After obtaining the target data, the target intelligent operator is input to obtain the first result.

[0112] For example, the target intelligence operator is a frame-in-frame detection operator, used to detect whether multiple targets exist simultaneously in the same frame of a single image or video. Based on target detection algorithms, it extracts image features through a deep learning model to achieve simultaneous detection and judgment of multiple targets. The input of the frame-in-frame detection operator is an image or video frame, and the output is a coexistence judgment, that is, the result of whether multiple targets exist simultaneously in the same frame. For example, if there cannot be more than 2 people in the same frame, the output is a judgment result of "whether more than 2 people exist". It can also output information such as the location and category of the targets.

[0113] For example, the target intelligent operator is a TTS operator. The TTS operator has a built-in speech synthesis model, such as an end-to-end neural network model. Text data is input into the TTS operator, which maps the text into the acoustic features of speech. Then, it is converted into an audio waveform by a vocoder and outputs the corresponding speech audio.

[0114] For example, the target intelligent operator is an ASR operator. The ASR operator has a built-in speech recognition model, such as an end-to-end multilingual speech recognition model. The speech audio is input into the ASR operator, which performs noise reduction, framing, feature extraction and other operations on the speech audio to identify phonemes and words in the speech, convert the continuous speech signal into a discrete text sequence, and perform grammatical correction and contextual semantic correction on the recognition result to output the corresponding text content.

[0115] After the target intelligent operator outputs the first result, it is transmitted to the central application via the RTSP protocol and / or Socket protocol.

[0116] In one possible embodiment, the business processing request is a frame detection request. After receiving this request, the central application starts the frame detection operator. The frame detection operator obtains video stream data through the central application, determines whether there are only 2 people in the business process, obtains the frame detection result, and returns the frame detection result to the central application. The central application transmits the frame detection result to the host computer to realize real-time quality inspection.

[0117] In one possible embodiment, the business processing request is a TTS request, and the text content is sent to the central application at the same time. After receiving this request and text content, the central application starts the TTS operator, which converts the text into speech and returns it to the central application. The central application transmits the speech to the host computer and calls the speaker module to start playing the speech.

[0118] In one possible embodiment, the service processing request is an ASR request. After receiving this request, the central application starts recording microphone data. The host application communicates with the central application to request the end of recording and to obtain the ASR result. The central application ends microphone recording and starts the ASR operator. The ASR operator calculates the recording data and outputs text to the central application. The central application then outputs the result to the host computer.

[0119] Specifically, the ASR function is implemented through microphone array signal processing technology and adaptive sound source localization technology. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a schematic diagram of a microphone array provided in an embodiment of this application, as shown below. Figure 4 As shown, a microphone array can include four microphones arranged in a straight line: a first microphone, a second microphone, a third microphone, and a fourth microphone. The timing and intensity at which these microphones receive the same sound signal will differ. For example, if the sound source is closer to the first microphone than to the fourth microphone, the sound will be received by the first microphone first. Therefore, beamforming algorithms can be used to adjust the weight and phase of the signals received by each microphone in the array, enabling the array to enhance sound signals from specific directions while suppressing sound signals from other directions. This allows for precise capture of the target sound and filtering of interfering sounds.

[0120] Among them, such as Figure 4 As shown, beams 0 (90°), 1 (30°), and 2 (150°) are set. Based on the set angles, the weight and phase of the signals received by each microphone are adjusted to regulate the microphone array's sensitivity to sounds from these directions. For example, the microphone array can amplify sound signals from the 90° direction and suppress sound signals from other directions.

[0121] As can be seen, in this embodiment, by setting the microphone array and the appropriate beam direction, the customer's voice can be focused, ambient noise and other people's conversations can be suppressed, making the recorded sound clearer.

[0122] Please refer to Figure 5 , Figure 5 This is a schematic diagram of a sound source localization scenario provided in an embodiment of this application, such as... Figure 5 As shown, the microphone array device uses self-localizing sound source technology to determine the pickup area and the suppression area. The pickup area refers to the direction and location of the sound source at the customer's location, while the suppression area refers to the area where ambient noise is located.

[0123] The process involves acquiring multi-channel raw speech collected by a microphone array, dividing the space around the microphone into a pickup area and a suppression area, matching the sound source location output by the self-localization sound source technology with the pickup area and suppression area, determining the region to which each sound source belongs, and labeling each sound source with a pickup area label or a suppression area label.

[0124] Specifically, the beamformer is optimized by designing the main lobe direction of the beamforming based on the location information of the target sound source within the pickup area, thereby maximizing the sound gain of the microphone array in that direction. For interfering sound sources within the suppression area, a null direction is designed using the beamforming algorithm to attenuate the location of the interfering sound source and reduce its signal strength.

[0125] The process involves inputting the original multi-channel speech signal into an optimized beamformer, adjusting the phase and weighting the amplitude of each channel signal, and then generating a single-channel enhanced signal through summation or weighted merging. At this point, the output signal highlights the target sound source in the pickup area while suppressing interference sound sources in the suppression area.

[0126] As can be seen, in this embodiment, the sound in the pickup area can be picked up with high quality, while the interference sound in the suppression area can be suppressed, thereby improving the quality of sound acquisition. This allows the voices of both the customer and the teller to be effectively recorded while suppressing environmental noise and interference sound as much as possible.

[0127] In one possible embodiment, the central application includes a software tool development kit. After obtaining the first result obtained by the target intelligent operator processing the target data, the method further includes: obtaining the target accuracy of the first result; detecting that the target accuracy is less than a first preset accuracy, and determining upgrade feedback information based on the target accuracy; sending the upgrade feedback information to the host computer; receiving upgrade data sent by the host computer through the software tool development kit; and performing an update operation on the target intelligent operator based on the upgrade data.

[0128] The first preset accuracy is a pre-set qualification standard, that is, the minimum accuracy standard that the calculation result quality must reach, such as the ASR recognition accuracy rate needing to be greater than or equal to 95%.

[0129] Each AI operator outputs a result, including a confidence level, which is then used as the target accuracy of the result.

[0130] In one possible embodiment, a corresponding quality assessment operator can be configured for each AI operator. The quality assessment operator is used to evaluate the quality of the first result to obtain the target accuracy. For example, for the frame-in-frame detection result, the corresponding quality assessment operator is used to determine the completeness of the frame and output the target accuracy. For example, if only half of a face is visible, the algorithm may misclassify it as a frame-in-frame detection.

[0131] The process involves comparing the target accuracy with a first preset accuracy. If the target accuracy is not met, an escalation feedback process is triggered. For example, if the first preset accuracy is 90% and the target accuracy is 89%, the result is deemed unqualified, and the escalation feedback process is triggered.

[0132] In one possible embodiment, determining the upgrade feedback information based on the target accuracy includes: determining the fluctuation range of the target accuracy; determining the performance index of the target intelligent operator in processing the target data; determining the upgrade range based on the fluctuation range and the performance index; and generating the upgrade feedback information based on the upgrade range.

[0133] The fluctuation range of target accuracy refers to the degree of change in the accuracy of the intelligent operator's output over a period of time. Specifically, by statistically analyzing the target accuracy over a certain period, the difference between the maximum and minimum values ​​is calculated to obtain the fluctuation value, and the ratio of the fluctuation value to the average value is calculated to obtain the fluctuation range.

[0134] Large fluctuations indicate poor operator stability, requiring priority investigation and upgrades; small fluctuations indicate that the operator is currently stable and adjustments can be postponed.

[0135] Among them, performance metrics are used to measure the efficiency and resource consumption of operators when processing data, and to judge the health of operators at the technical level.

[0136] Specifically, the processing speed index is obtained by calculating the amount of data processed per second and the average response time; the resource consumption index is obtained by calculating the CPU utilization, image processor utilization, and memory usage; and the robustness index is obtained by calculating the accuracy maintenance capability of the AI ​​operator under abnormal data such as blurred images and noisy speech.

[0137] Specifically, when the fluctuation range is small and the performance indicators are slightly abnormal, the corresponding upgrade scope is a partial upgrade, which only requires adjusting some parameters or configurations of the AI ​​operator. For example, if the fluctuation range is within the preset threshold and the performance indicators slightly deviate from the standard, such as an increase in response time of 0.2 seconds, without affecting the overall functionality, then a partial upgrade is appropriate.

[0138] When the fluctuation exceeds the preset threshold and the performance indicators are poor, the corresponding upgrade scope is a global upgrade, which requires replacing the entire AI operator.

[0139] This involves translating the upgrade scope into specific, actionable upgrade feedback information for development or operations personnel to reference, ensuring the upgrade's successful implementation. Feedback information can include problem location, fluctuation range, and specific anomalies in performance metrics; it can also include upgrade goals, i.e., the effects to be achieved after the upgrade; and it can include priorities, indicating the degree of urgency.

[0140] As can be seen, in this embodiment, by determining the upgrade scope and generating upgrade feedback information based on the fluctuation range of target accuracy and the performance indicators of the target intelligent operator in processing target data, the upgrade of the intelligent operator can be made more precise and efficient. This avoids the waste of resources caused by blind upgrades, addresses core issues in a targeted manner, ensures the stability of the system during the upgrade process, and reduces upgrade risks. Simultaneously, it can promote the continuous evolution of operator performance, enhance system controllability, ensure business continuity, and thus increase customer trust in the system. Ultimately, it finds a balance between problem-solving and cost control, enabling the intelligent operator to better adapt to business scenario requirements.

[0141] The upgrade feedback information is sent to the host computer via a serial port protocol. The host computer then retrieves and installs the latest update program based on the feedback information and initiates the update process. This update program calls the software development kit (SDK) of the device's internal central application to transmit the upgrade data to the central application.

[0142] In one possible embodiment, performing an update operation on the target intelligent operator based on the upgrade data includes: if the upgrade data is detected to be parameter update data for the target intelligent operator, then updating the parameters of the target intelligent operator in the operator container based on the parameter update data.

[0143] Partial upgrades involve smaller changes, affecting only certain parts of the operator, such as configuration files and model parameters. The upgrade data includes update packages and update scripts. The update packages contain information about files to be overwritten, files to be deleted, and files to be modified.

[0144] Different AI operators correspond to different update scripts.

[0145] Upon receiving the update command, the central application locates the target AI operator in the operator container based on its ID, and executes the update script from the upgrade data to complete the update. After the script finishes execution, the central application deletes the update package and update script used for this update, and then notifies the update program that the upgrade is complete.

[0146] In one possible embodiment, the step of performing an update operation on the target smart operator based on the upgrade data includes: if the upgrade data is detected to be the image data to be updated of the target smart operator, then deleting the target smart operator in the operator container; and deploying the image in the operator container based on the image data to be updated.

[0147] The global upgrade requires replacing the entire AI operator. The update program uses the software development kit (SDK) of the central application within the device to transmit the image of the AI ​​operator to be updated to the central application via a serial port protocol. After transmission, the central application locates the target AI operator in the operator container based on its ID. If found, it uninstalls and deletes the target AI operator. Simultaneously, the central application executes an image deployment command to deploy the image of the new AI operator into the operator container. After deployment, the central application deletes the image of the AI ​​operator used in this update and then notifies the update program that the upgrade is complete.

[0148] As can be seen, in this embodiment of the application, different upgrade processes and upgrade ranges are configured for different AI operators to achieve precise upgrades of intelligent operators.

[0149] In one possible embodiment, AI operators can be expanded in the operator container according to customer needs. By adding functional units with AI processing capabilities, the intelligent processing capabilities can be expanded, and the system can be deployed and run under the premise of resource adaptation.

[0150] Within the limits of computing power, the AI ​​operators to be expanded are deployed dynamically, taking into account the resource management capabilities of the operator container and business needs. The operator container monitors computing resources in real time, and when resources are sufficient, it deploys the image and related parameters of the newly added AI operator according to business priorities. The deployment process is the same as the global upgrade process, and will not be described in detail here.

[0151] In one possible embodiment, the method further includes: detecting that the target accuracy is less than a second preset accuracy, then determining at least one reference intelligent operator in the operator container based on the target intelligent operator and the business processing task, wherein the second preset accuracy is greater than the first preset accuracy; constructing a business processing model based on the target intelligent operator and the at least one reference intelligent operator; transmitting the target data to the business processing model through the first communication protocol and / or the second communication protocol; obtaining a second result obtained by the business processing model processing the target data through the first communication protocol and / or the second communication protocol; and transmitting the second result to the host computer.

[0152] When the target accuracy is lower than the second preset accuracy, it means that the current target AI operator needs to be synchronously calculated by the auxiliary operator in order to improve the accuracy of the business processing task.

[0153] Specifically, operators that are related to the target intelligent operator can be selected from the operator container. Then, based on the current task type, reference AI operators that can help improve the results can be selected from the related operators. These reference AI operators must match the business task and have the ability to compensate for the deficiencies of the target intelligent operator.

[0154] This involves combining the target intelligent operator with a reference intelligent operator to form a collaborative business processing model. For example, the reference intelligent operator optimizes the output of the target intelligent operator to produce the final result.

[0155] In one possible embodiment, constructing a business processing model based on the target intelligent operator and the at least one reference intelligent operator includes: determining a first algorithm framework for the target intelligent operator; and determining at least one second algorithm framework for the at least one reference intelligent operator; determining the data dependency relationship between the target intelligent operator and the at least one reference intelligent operator based on the first algorithm framework and the at least one second algorithm framework; acquiring the computing power resources of the target intelligent operator and the at least one reference intelligent operator; and constructing the business processing model based on the data dependency relationship and the computing power resources.

[0156] The first algorithm framework includes the algorithmic logic, input data requirements, and output data format of the target AI operator. The second algorithm framework includes the algorithmic logic and input / output specifications of each reference AI operator. Data dependencies between operators can be derived from the algorithm framework to determine whether the output of a preceding operator matches the input of a subsequent operator. For example, before starting the ASR operator, the denoising operator can be started first, and its output can be input into the ASR operator.

[0157] In one possible embodiment, to coordinate computing power, a resource reservation mechanism is set up, pre-allocating 15%-20% of the computing power buffer for high-priority and long-duration AI operators. Furthermore, the priorities of AI operators are dynamically set, adjusting them according to their latency sensitivity. Simultaneously, a micro-task aggregation strategy is implemented, placing short-latency AI operators into a task queue for batch processing.

[0158] First, the execution order is initially determined based on data dependencies. Then, for AI operators without data dependencies, the execution order is optimized using computing resources.

[0159] If multiple operators have no data dependencies and sufficient computing power, they can be executed in parallel; if computing power is insufficient, they can be executed serially.

[0160] The integrated model includes both the execution order of operators and the resource allocation scheme for each operator, enabling it to run directly in a real-world environment.

[0161] Specifically, the target data is transmitted to the business processing model via the RTSP protocol and / or the Socket protocol. After receiving the target data, the model processes the target data according to the execution order of the intelligent operators, and finally generates a second result. Then, the second result is output to the central application via the RTSP protocol and / or the Socket protocol. The central application transmits the second result to the host computer via the serial port protocol.

[0162] As can be seen, in this embodiment of the application, by coordinating the data dependencies and computing power among operators, different operators can complement each other, thereby improving the accuracy of business processing.

[0163] S250, the first result is transmitted to the host computer.

[0164] If the detected target accuracy is greater than or equal to the second preset accuracy, the first result is transmitted to the host computer via serial port protocol.

[0165] As can be seen, in this embodiment, the operator container pre-deploys various intelligent operator images into the dual-recording device. The central application transmits the acquired target data to the target intelligent operator for processing through different communication methods, and returns the processing result to the central application. This enables the dual-recording service to achieve efficient data processing and business logic execution through the collaborative architecture of operator containerization deployment, flexible communication mechanisms, and intelligent processing capabilities. Furthermore, the localized response processing of business processing requests improves data security and avoids processing delays caused by network fluctuations or excessive load. It also simplifies the business processing flow, thereby improving business response speed and business processing efficiency.

[0166] In one possible embodiment, the dual-recording device supports simultaneous operation of two cameras, switching between the primary and secondary cameras, and achieving dynamic picture-in-picture compositing.

[0167] In one possible embodiment, the size of the picture-in-picture can be dynamically adjusted based on a dynamic detection algorithm. Specifically, when the dynamic detection algorithm detects movement of an object on the main monitoring screen, the picture-in-picture will automatically shrink; when the dynamic detection shows that the main screen is in a low-dynamic or static state, such as when no object enters the area, the picture-in-picture will appropriately enlarge.

[0168] In one possible implementation, a size adjustment may be triggered when dynamic detection detects a correlation between the content of the main screen and the picture-in-picture. For example, if the target object tracked by the main screen moves near the area covered by the picture-in-picture, the picture-in-picture will temporarily shrink to avoid information confusion caused by the overlap of the two contents; if the target object moves away from the picture-in-picture area, the picture-in-picture will restore its original size to ensure that the auxiliary information is not over-compressed.

[0169] In one possible embodiment, the acoustic structure of the dual recording device is an integrated sealed cavity design.

[0170] In one possible embodiment, the dual-recording device can be powered by a single USB 2.0 cable, enabling the simultaneous operation of dual cameras, speakers, microphone arrays, and LED indicators.

[0171] In one possible embodiment, the dual-recording device has voice wake-up and voice dialogue functions, and constructs a complete link from sound acquisition to command response through the coordinated operation of microphone array, AI noise reduction application, neural network processor inference thread and local command dictionary.

[0172] Specifically, when a user utters a wake-up word, the voice signal collected by the microphone array first enters the AI ​​noise reduction application, where the audio is further purified through deep learning algorithms. The purified voice signal is then sent to the neural network processor inference thread to efficiently run the voice wake-up model and the voice recognition model.

[0173] During the wake-up phase, the processed audio features are compared with the wake-up word features pre-stored in the local command dictionary. Once a match is found, the device switches from sleep mode to working mode, ready to receive subsequent dialogue commands.

[0174] During the dialogue phase, the neural network processor continues to analyze the user's voice content in real time, converting continuous speech into text information and matching it with commonly used commands in the command dictionary to quickly generate corresponding execution instructions, driving the device to complete the corresponding operation. This enables accurate wake-up of dual-recording devices, efficiently understanding and executing user voice commands, and achieving a natural and smooth voice interaction experience.

[0175] For examples consistent with the above embodiments, please refer to... Figure 6 , Figure 6 This is a functional unit block diagram of a service processing device for a dual-recording device provided in an embodiment of this application, such as... Figure 6 As shown, the service processing device 60 of the dual-recording device includes: a receiving unit 61, configured to receive a service processing request from the host computer, the service processing request including a service processing task; a determining unit 62, configured to acquire target data according to the service processing task via a first communication protocol and / or a second communication protocol; and to determine a target intelligent operator in the operator container according to the service processing task; a first transmission unit 63, configured to transmit the target data to the target intelligent operator via the first communication protocol and / or the second communication protocol; an acquiring unit 64, configured to acquire a first result obtained by the target intelligent operator processing the target data via the first communication protocol and / or the second communication protocol; and a second transmission unit 65, configured to transmit the first result to the host computer.

[0176] In one possible embodiment, the central application includes a software tool development kit. After obtaining the first result obtained by the target intelligent operator processing the target data, the service processing device 60 of the dual-recording device is further configured to: obtain the target accuracy of the first result; detect that the target accuracy is less than a first preset accuracy, and determine upgrade feedback information based on the target accuracy; send the upgrade feedback information to the host computer; receive upgrade data sent by the host computer through the software tool development kit; and perform an update operation on the target intelligent operator based on the upgrade data.

[0177] In one possible embodiment, in determining the upgrade feedback information based on the target accuracy, the service processing device 60 of the dual-recording device is further configured to: determine the fluctuation range of the target accuracy; determine the performance index of the target intelligent operator in processing the target data; determine the upgrade range based on the fluctuation range and the performance index; and generate the upgrade feedback information based on the upgrade range.

[0178] In one possible embodiment, in performing an update operation on the target intelligent operator based on the upgrade data, the service processing device 60 of the dual recording device is further configured to: if the upgrade data is detected to be parameter update data of the target intelligent operator, update the parameters of the target intelligent operator in the operator container based on the parameter update data.

[0179] In one possible embodiment, in performing an update operation on the target intelligent operator based on the upgrade data, the service processing device 60 of the dual-recording device is further configured to: if the upgrade data is detected to be the image data to be updated of the target intelligent operator, delete the target intelligent operator in the operator container; and perform image deployment in the operator container based on the image data to be updated.

[0180] In one possible embodiment, the service processing device 60 of the dual-recording device is further configured to: detect that the target accuracy is less than a second preset accuracy, then determine at least one reference intelligent operator in the operator container according to the target intelligent operator and the service processing task, wherein the second preset accuracy is greater than the first preset accuracy; construct a service processing model according to the target intelligent operator and the at least one reference intelligent operator; transmit the target data to the service processing model through the first communication protocol and / or the second communication protocol; obtain a second result obtained by the service processing model from processing the target data through the first communication protocol and / or the second communication protocol; and transmit the second result to the host computer.

[0181] In one possible embodiment, in constructing a service processing model based on the target intelligent operator and the at least one reference intelligent operator, the service processing apparatus 60 of the dual-recording device is further configured to: determine a first algorithm framework for the target intelligent operator; and determine at least one second algorithm framework for the at least one reference intelligent operator; determine the data dependency relationship between the target intelligent operator and the at least one reference intelligent operator based on the first algorithm framework and the at least one second algorithm framework; acquire the computing power resources of the target intelligent operator and the at least one reference intelligent operator; and construct the service processing model based on the data dependency relationship and the computing power resources.

[0182] It is understood that since the method embodiments and the device embodiments are different presentations of the same technical concept, the content of the method embodiment section in this application should be adapted to the device embodiment section in a synchronous manner, and will not be repeated here.

[0183] In the case of using integrated units, please refer to Figure 7 , Figure 7 This is a functional unit block diagram of another dual-recording device's service processing apparatus provided in this application embodiment, such as... Figure 7As shown, the service processing device 60 of the dual-recording device includes a processing module 602 and a communication module 601. The processing module 602 controls and manages the operation of the service processing device 60, for example, executing the steps of the receiving unit 61, the determining unit 62, the first transmission unit 63, the acquiring unit 64, and the second transmission unit 65, and / or performing other processes of the technology described herein. The communication module 601 is used for interaction between the service processing device 60 and other devices.

[0184] Among them, such as Figure 7 As shown, the service processing device 60 of the dual recording device may further include a storage module 603, which is used to store the program code and data of the service processing device 60 of the dual recording device.

[0185] The processing module 602 may be a processor or controller, such as a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor may also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0186] The communication module 601 can be a transceiver, RF circuit, or communication interface, etc. The storage module 603 can be a memory.

[0187] All relevant content in each scenario involved in the above method embodiments can be referenced from the functional descriptions of the corresponding functional modules, and will not be repeated here. The business processing device 60 of the above dual-recording device can execute the above... Figure 3 The business processing method of the dual recording device shown.

[0188] Please see Figure 8 , Figure 8 This is a schematic diagram of the structure of a dual-recording device proposed in an embodiment of this application, as shown below. Figure 8As shown, the dual recording device 102 includes a processor 810, a memory 820, a communication interface 830, and one or more programs 821. The one or more programs 821 are stored in the memory and configured to be executed by the processor. When the program is executed, it includes some or all of the steps of the business processing method of any dual recording device described in the above method embodiments. The processor, memory, and communication interface are interconnected and complete communication between them.

[0189] The memory can be volatile memory such as Dynamic Random Access Memory (DRAM) or non-volatile memory such as a hard disk drive. The memory stores a set of executable program code, and the processor calls the executable program code stored in the memory to execute some or all of the steps of any of the business processing methods of the dual-recording device described in the above embodiments of the business processing method for the dual-recording device.

[0190] As can be seen, the dual-recording device 102 described in this application embodiment first receives a service processing request from the host computer, the service processing request including a service processing task; then, according to the service processing task, it obtains target data through a first communication protocol and / or a second communication protocol; and determines a target intelligent operator in the operator container according to the service processing task; then, it transmits the target data to the target intelligent operator through the first communication protocol and / or the second communication protocol; then, it obtains a first result obtained by the target intelligent operator processing the target data through the first communication protocol and / or the second communication protocol; finally, it transmits the first result to the host computer.

[0191] This application's operator container pre-deploys various intelligent operator images into the dual-recording device. The central application transmits the acquired target data to the target intelligent operator for processing through different communication methods, and returns the processing results to the central application. This enables the dual-recording service to achieve efficient data processing and business logic execution through the collaborative architecture of operator containerization deployment, flexible communication mechanisms, and intelligent processing capabilities. Furthermore, by handling business processing requests locally, data security is improved, and processing delays caused by network fluctuations or excessive load are avoided. At the same time, the business processing flow is simplified, thereby improving business response speed and business processing efficiency.

[0192] This application also provides a computer storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the methods described in the above method embodiments, wherein the computer includes a dual recording device.

[0193] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments. The computer program product may be a software installation package, and the computer may include a dual-recording device.

[0194] It should be noted that, for the sake of simplicity, the aforementioned methods are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are optional, and the actions and modules involved are not necessarily essential to this application.

[0195] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0196] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0197] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0198] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software program module.

[0199] If the integrated unit is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0200] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: a flash drive, a read-only memory, a random access memory, a magnetic disk, or an optical disk, etc.

[0201] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A service processing method for a dual-recording device, characterized in that, A central application is applied to a dual-recording device. The central application includes a software development kit (SDK). The dual-recording device also includes an operator container, which contains multiple categories of intelligent operators. Each category of intelligent operator is used to process a single dual-recording service. Images of the multiple categories of intelligent operators are pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer. The method includes: Receive a service processing request from the host computer, the service processing request including a service processing task; According to the business processing task, target data is acquired through a first communication protocol and / or a second communication protocol; and, according to the business processing task, a target intelligent operator is determined in the operator container. The target data is transmitted to the target intelligent operator through the first communication protocol and / or the second communication protocol; The system obtains a first result from the target intelligent operator processing the target data using the first communication protocol and / or the second communication protocol; wherein, the target accuracy of the first result is obtained; if the target accuracy is detected to be less than a first preset accuracy, upgrade feedback information is determined based on the target accuracy; the upgrade feedback information is sent to the host computer; upgrade data sent by the host computer is received through the software tool development kit; an update operation is performed on the target intelligent operator based on the upgrade data; wherein, if the target accuracy is detected to be less than a second preset accuracy, at least one reference intelligent operator is determined in the operator container based on the target intelligent operator and the business processing task, and the second preset accuracy is greater than the first preset accuracy; a business processing model is constructed based on the target intelligent operator and the at least one reference intelligent operator; the target data is transmitted to the business processing model through the first communication protocol and / or the second communication protocol; a second result from the business processing model processing the target data is obtained through the first communication protocol and / or the second communication protocol; the second result is transmitted to the host computer. The first result is transmitted to the host computer.

2. The method according to claim 1, characterized in that, The step of determining the upgrade feedback information based on the target accuracy includes: Determine the fluctuation range of the target accuracy; Determine the performance metrics of the target intelligent operator in processing the target data; The upgrade range is determined based on the fluctuation range and the performance indicators; The upgrade feedback information is generated based on the upgrade scope.

3. The method according to claim 1, characterized in that, The step of performing an update operation on the target intelligent operator based on the upgrade data includes: If the upgrade data is detected to be parameter update data for the target intelligent operator, then the parameters of the target intelligent operator are updated in the operator container according to the parameter update data.

4. The method according to claim 1, characterized in that, The step of performing an update operation on the target intelligent operator based on the upgrade data includes: If the upgrade data is detected to be the image data to be updated of the target smart operator, then the target smart operator is deleted from the operator container; Based on the image data to be updated, the image is deployed in the operator container.

5. The method according to claim 1, characterized in that, The step of constructing a business processing model based on the target intelligent operator and the at least one reference intelligent operator includes: A first algorithm framework for the target intelligent operator is determined; and at least one second algorithm framework for the at least one reference intelligent operator is determined. Based on the first algorithm framework and the at least one second algorithm framework, the data dependency relationship between the target intelligent operator and the at least one reference intelligent operator is determined; Acquire the computing power resources of the target intelligent operator and the at least one reference intelligent operator; Based on the data dependencies and computing resources, the business processing model is constructed.

6. A service processing device for a dual-recording equipment, characterized in that, A central application for dual-recording devices, the dual-recording device further includes an operator container, the operator container including multiple categories of intelligent operators, a single category of intelligent operator for processing a single dual-recording service, and images of the multiple categories of intelligent operators pre-deployed in the operator container. The dual-recording device is communicatively connected to a host computer. The device includes: A receiving unit is configured to receive a service processing request from the host computer, wherein the service processing request includes a service processing task. The determining unit is configured to acquire target data according to the business processing task via a first communication protocol and / or a second communication protocol; and to determine a target intelligent operator in the operator container according to the business processing task. The first transmission unit is used to transmit the target data to the target intelligent operator through a first communication protocol and / or a second communication protocol; The acquisition unit is configured to acquire, through a first communication protocol and / or a second communication protocol, the first result obtained by the target intelligent operator processing the target data; The second transmission unit is used to transmit the first result to the host computer; The central application includes a software tool development kit. After obtaining the first result obtained by the target intelligent operator processing the target data, the device further includes: obtaining the target accuracy of the first result; detecting that the target accuracy is less than a first preset accuracy, then determining upgrade feedback information based on the target accuracy; sending the upgrade feedback information to the host computer; receiving upgrade data sent by the host computer through the software tool development kit; performing an update operation on the target intelligent operator based on the upgrade data; detecting that the target accuracy is less than a second preset accuracy, then determining at least one reference intelligent operator in the operator container based on the target intelligent operator and the business processing task, wherein the second preset accuracy is greater than the first preset accuracy; constructing a business processing model based on the target intelligent operator and the at least one reference intelligent operator; transmitting the target data to the business processing model through the first communication protocol and / or the second communication protocol; obtaining the second result obtained by the business processing model processing the target data through the first communication protocol and / or the second communication protocol; and transmitting the second result to the host computer.

7. A dual-recording device, characterized in that, The device includes: The device includes a memory, a processor, and executable program code stored in the memory and executable on the processor, wherein the processor executes the executable program code to perform the steps of the service processing method of the dual recording device as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores executable program code, which includes execution instructions for performing the steps of the service processing method of the dual-recording device as described in any one of claims 1-5.

Images