Driving test video generation system and method

Abstract

The invention provides a driving test video generation system and method, and belongs to the technical field of video processing, and the system comprises a driving test data collection module deployed at an end side device, a video generation module and a resource management module driven by an end side AI, and a video generation module deployed at a cloud side AI driven by a cloud side device. The driving test data acquisition module acquires driving test data of a plurality of data sources in a driving test process; analyzing the driving test data by a video generation module driven by an end side AI, and determining a corresponding video template type; the resource management module is used for generating a video generation module driven by a target AI according to the type of the video template and the processing capacity of the end side equipment; the target AI-driven video generation module generates the target driving test video based on the driving test data and the video template type, the AI-driven generation function is utilized, professional editing skills are not needed, the video can be generated in a zero-editing mode, and meanwhile the end side device and the cloud side device are utilized for cooperative work, so that the video generation efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of video processing technology, specifically to a driving test video generation system and method. Background Technology

[0002] In the current field of driver's license test video creation and sharing, the technological shortcomings are mainly reflected in two aspects. First, the traditional video creation process relies on complex editing software, requiring users to manually perform operations such as editing, adding subtitles, and special effects. These operations are not only tedious but also require a high level of professional skills, increasing the difficulty for driver's license test users to quickly share their practice results and exam scores. Second, because video data processing consumes a lot of computing resources, the use of a single cloud storage and processing model reduces the efficiency of video generation.

[0003] Therefore, in order to improve the efficiency of driving test video generation, it is particularly important to develop a system that can simplify the video creation process and optimize it for driving test scenarios. Summary of the Invention In view of this, it is necessary to provide a driving test video generation system and method to solve the technical problems of low accuracy and low recognition efficiency in the existing driving test video generation technology.

[0004] To address the aforementioned technical problems, in a first aspect, the present invention provides a driver's test video generation system, comprising: a driver's test data acquisition module, an edge AI-driven video generation module, and a resource management module deployed on a terminal device, and a cloud AI-driven video generation module deployed on a cloud device, wherein the processing power of the cloud AI-driven video generation module is higher than that of the edge AI-driven video generation module. The driving test data acquisition module is used to collect driving test data from multiple data sources during the driving test process and send the driving test data to the AI-driven video generation module on the edge. The edge AI-driven video generation module is used to parse the driving test data and determine the corresponding video template type; The resource management module is used to determine the target AI-driven video generation module as either the edge AI-driven video generation module or the cloud AI-driven video generation module based on the video template type and the processing capability of the edge device. The target AI-driven video generation module is used to generate target driving test videos based on the driving test data and the video template type.

[0005] In one possible implementation, the end-side device is further deployed with a hierarchical encryption module, which includes a hierarchical unit and an encryption unit. The grading unit is used to grade the multi-source driver's test data according to the privacy level of the data, and to determine the encryption level of driver's test data from different data sources. The encryption unit includes multiple encryption sub-units, each of which corresponds to an encryption level. The encryption sub-unit is used to encrypt the corresponding driving test data according to the encryption level and send the encrypted driving test data to the cloud-side device.

[0006] In one possible implementation, the edge device stores a lightweight template library, and the cloud device stores a full-featured template library; the resource management module is specifically used for: If the processing capability of the edge device meets the preset resource processing conditions and the lightweight template library contains a template library that matches the video template type, the edge AI-driven video generation module is determined to be the target AI-driven video generation module. If the processing capability of the edge device is not met by the preset resource processing conditions, or if the lightweight template library does not have a template library that matches the video template type, the cloud-side AI-driven video generation module is determined to be the target AI-driven video generation module.

[0007] In one possible implementation, when the target AI-driven video generation module is an edge AI-driven video generation module, the edge AI-driven video generation module is used for: A template matching the video template type is selected from the lightweight template library to obtain a target video template, and the driving test feature information corresponding to the driving test data is filled into the target video template to generate a target driving test video; When the target AI-driven video generation module is a cloud-based AI-driven video generation module, the cloud-based AI-driven video generation module is used for: After selecting a template that matches the video template type from the full-quantization template library to obtain the target video template, the driving test feature information corresponding to the driving test data is filled into the target video template to generate the target driving test video, and the target driving test video is sent to the end device.

[0008] In one possible implementation, the cloud-side device is also deployed with a security sandbox; the cloud-side AI-driven video generation module is also used for: Receive encrypted driving test data corresponding to the driving test data transmitted to the security sandbox via HTTPS protocol; Based on the encrypted driving test data and the fully quantized template, a target driving test video is generated in the secure sandbox. The target driving test video is blurred and encrypted using differential privacy technology, and then returned to the end device.

[0009] In one possible implementation, the cloud-side device is further deployed with a blockchain evidence storage module; the cloud-side AI-driven video generation module is also used for: The transmission and processing of the encrypted driver's test data are uploaded to the blockchain evidence storage module.

[0010] In one possible implementation, the cloud-side device is further deployed with a template library update module, which includes a template scoring unit, a template visual element update unit, and a user-driven template update unit. The template scoring unit is used to periodically calculate the comprehensive score of each template in the full-quantification template library; The template visual element replacement unit is used to replace visual elements in templates with a comprehensive score less than a preset score threshold with target visual elements to generate subsequent templates. The target visual elements are visual elements extracted from templates of the same type with a comprehensive score greater than or equal to the preset score threshold by an AI element design model. The user-driven template update unit is used to obtain feedback data from target users regarding the use of the candidate templates within a preset number of periods, and to select an update template from the candidate templates based on the feedback data.

[0011] In one possible implementation, the edge AI-driven video generation module includes an edge AI parsing engine, a template decision unit, and a video synthesis unit; The edge AI analysis engine is used to extract driving test feature information from the driving test data; The template decision unit is used to analyze the driving test feature information and determine the target video template; The video synthesis unit is used to fill the driving test feature information into the target video template to generate the target driving test video.

[0012] In one possible implementation, the end-side device is further equipped with a video preview unit and a video forwarding unit; The video preview unit is used to display the received target driving test video; The video forwarding unit is used to forward the received target driving test video to a social media platform.

[0013] Secondly, the present invention also provides a method for generating driver's license test videos, applied to the driver's license test video generation system described in the first aspect, comprising: Acquire driving test data from multiple data sources during the driving test process; The driving test data is analyzed to determine the corresponding video template type; Based on the video template type and the processing capability of the edge device, a target AI-driven video generation module is determined, wherein the target AI-driven video generation module is either an edge AI-driven video generation module or a cloud AI-driven video generation module. Based on the driving test data and the video template type, the target driving test video is generated by the target AI-driven video generation module.

[0014] The beneficial effects of this invention are: The driving test video generation system provided by this invention includes a driving test data acquisition module, an edge AI-driven video generation module, and a resource management module deployed on a terminal device, as well as a cloud AI-driven video generation module deployed on a cloud device. The cloud AI-driven video generation module has a higher processing power than the edge AI-driven video generation module. The driving test data acquisition module is used to collect driving test data from multiple data sources during the driving test process and send the driving test data to the edge AI-driven video generation module to ensure the comprehensiveness and real-time nature of the collected driving test data. The edge AI-driven video generation module is used to parse the driving test data, determine the corresponding video template type, and through real-time parsing on the terminal device, change the video template matching of the driving test scenario from "cloud waiting" to "local second-level," which can improve the processing of video data by utilizing the integrated perception-understanding-generation capabilities of AI. The system analyzes and generates video to improve efficiency. A resource management module determines either the edge-side AI-driven video generation module or the cloud-side AI-driven video generation module as the target AI-driven video generation module based on the video template type and the processing capabilities of the edge device. The edge and cloud devices work collaboratively to achieve a cyclical process of edge-first, cloud-backup, and dynamic switching, thus improving video generation efficiency. The target AI-driven video generation module generates a target driving test video based on the driving test data and the video template type. Utilizing AI-driven generative capabilities, it generates videos with zero editing required, eliminating the need for professional editing skills. Simultaneously, it fully leverages the collaborative work of edge and cloud devices to achieve intelligent task scheduling between the edge and cloud sides, improving video generation efficiency and meeting the needs of driving test users for convenient and secure sharing of driving test videos on social networks. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic flowchart of an embodiment of the driver's test video generation system provided by the present invention; Figure 2 The template dynamic update flowchart provided by this invention; Figure 3 The zero-edit video generation flowchart provided by this invention; Figure 4 This is an architecture diagram of the driver's test data zero-edit video generation system based on privacy-first edge-cloud collaboration provided by the present invention; Figure 5 This is a diagram of the privacy-first cloud collaboration architecture provided by the present invention; Figure 6 This is a schematic flowchart of an embodiment of the driver's test video generation method provided by the present invention; Figure 7 The timing diagram for edge-cloud task scheduling provided by this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0018] In the description of the embodiments of the present invention, unless otherwise stated, "a plurality of" means two or more.

[0019] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0020] 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 the invention. 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.

[0021] This invention provides a driving test video generation system and method, which will be described below.

[0022] Figure 1 A schematic diagram illustrating the structure of a specific embodiment of the driver's test video generation system provided by the present invention is shown.

[0023] The driving test video generation system includes: a driving test data acquisition module 101, an edge AI-driven video generation module 102, and a resource management module 103 deployed on the edge device 100, and a cloud AI-driven video generation module 201 deployed on the cloud device 200. The processing power of the cloud AI-driven video generation module 201 is higher than that of the edge AI-driven video generation module 102. The driving test data acquisition module 101 is used to collect driving test data from multiple data sources during the driving test process and send the driving test data to the AI-driven video generation module 102 on the edge.

[0024] The driver's license test data acquisition module 101 is configured to collect driver's license test data in real time from multiple data sources, including on-board sensor units, test scoring terminals, and / or user interaction interfaces, during the driver's license test process. For example, a data acquisition unit is set up on the user terminal device (such as a smartphone or tablet) to collect multi-source data during the driver's license test process, including test score data (subject name, score, ranking, deduction items, etc.), practice operation data (practice time for each item, number of operation errors, operation speed change curve, etc.), and vehicle driving trajectory data (driving route, steering angle, speed fluctuation, etc.), to ensure the comprehensiveness and real-time nature of the collected driver's license test data.

[0025] The edge AI-driven video generation module 102 refers to a lightweight AI subsystem deployed on the edge device 100, such as a user terminal (mobile phone / tablet / vehicle system). This subsystem includes MobileNet-based OCR models and LSTM time-series analysis models. It performs real-time analysis of the collected driving test data and extracts key information. For example, it automatically recognizes the scores on the test score sheet through the OCR model and analyzes the progress trend of the operation data using the LSTM model. This enables the generation of driving test videos with zero editing, including data parsing, template matching, and local rendering. This allows the GPU of the edge device 1100 to be called to complete the video synthesis locally. The entire process does not require manual editing by the user, achieving "zero-edit" video generation.

[0026] The cloud-side AI-driven video generation module 201 is an AI computing cluster with higher processing power than the edge-side AI-driven video generation module 102. It is an AI computing cluster specifically designed for high-complexity tasks such as high-definition rendering, large-scale template retrieval, and complex special effects compositing that edge devices cannot physically perform. It can handle more complex and computationally intensive video generation tasks.

[0027] The edge AI-driven video generation module 101 is used to parse the driving test data and determine the corresponding video template type.

[0028] Specifically, the edge-side AI-driven video generation module 101 analyzes the driving test data to determine the corresponding video template type. By analyzing the data in real time on the edge device, the matching of video templates for driving test scenarios is transformed from "waiting in the cloud" to "local instantaneous matching," while minimizing traffic and power consumption costs. This facilitates a user experience with zero waiting time, zero uploading, and zero barriers to entry, thereby improving the efficiency of subsequent video generation. Understandably, since both the edge-side AI-driven video generation module 101 and the cloud-side AI-driven video generation module 201 are AI-driven video generation modules, they can leverage AI's integrated perception-understanding-generation capabilities to improve the analysis and generation of video data, thereby enhancing video generation efficiency.

[0029] The resource management module 103 is used to determine the edge AI-driven video generation module 102 or the cloud AI-driven video generation module 201 as the target AI-driven video generation module based on the video template type and the processing capability of the edge device 100.

[0030] Processing capacity refers to the computing and analysis capabilities of the end-side device, such as based on CPU utilization, network bandwidth, etc.

[0031] The resource management module 103 is a lightweight system-level daemon deployed on the edge device 100. It includes three engines: a resource detection engine, which is used to scan the CPU utilization and network bandwidth of the edge device 100, and to determine whether the edge device 100 has a matching template based on the video template type; and a decision engine, which is used to select the edge AI-driven video generation module or the cloud AI-driven video generation module as the target AI-driven video generation module based on the detection results of the resource detection engine.

[0032] Specifically, the resource management module 103 can intelligently decide whether to use the edge AI-driven video generation module or the cloud AI-driven video generation module to perform the video generation task based on the specific requirements of the current target video template and the actual processing capabilities of the edge device. In other words, it enables the edge device 100 and the cloud device 200 to work together, thereby realizing a loop process of edge priority, cloud backup, and dynamic switching, which improves video generation efficiency.

[0033] The target AI-driven video generation module is used to generate target driving test videos based on the driving test data and the video template type.

[0034] Specifically, the selected target AI-driven video generation module will generate videos based on driving test data and determined video template types, using AI-driven generative functions. No professional editing skills are required, and videos can be generated with zero editing. At the same time, it will make full use of the collaboration between edge devices and cloud devices to achieve intelligent task scheduling between edge and cloud, improve video generation efficiency, and meet the needs of driving test users to conveniently and safely share driving test videos on social networks.

[0035] In summary, the driving test video generation system provided in this embodiment of the invention includes a driving test data acquisition module, an edge AI-driven video generation module, and a resource management module deployed on a terminal device, as well as a cloud AI-driven video generation module deployed on a cloud device. The cloud AI-driven video generation module has a higher processing power than the edge AI-driven video generation module. The driving test data acquisition module is used to collect driving test data from multiple data sources during the driving test process and send the driving test data to the edge AI-driven video generation module to ensure the comprehensiveness and real-time nature of the collected driving test data. The edge AI-driven video generation module is used to parse the driving test data, determine the corresponding video template type, and through real-time parsing on the terminal device, change the video template matching of the driving test scenario from "cloud waiting" to "local second-level," which can utilize the integrated perception-understanding-generation capabilities of AI to improve the accuracy of video matching. The system enhances video generation efficiency through the analysis and generation capabilities of video data. A resource management module determines either the edge-side AI-driven video generation module or the cloud-side AI-driven video generation module as the target AI-driven video generation module based on the video template type and the processing capabilities of the edge device. The edge device and cloud device work collaboratively, enabling a cyclical process of edge-first, cloud-backup, and dynamic switching, thus improving video generation efficiency. The target AI-driven video generation module generates target driving test videos based on the driving test data and the video template type. Utilizing AI-driven generative functions, it allows for zero-edit video generation without requiring professional editing skills. Simultaneously, it fully leverages the collaborative work of edge and cloud devices to achieve intelligent task scheduling between the edge and cloud sides, improving video generation efficiency and meeting the needs of driving test users for convenient and secure sharing of driving test videos on social networks.

[0036] In some embodiments of the present invention, the end-side device is further deployed with a hierarchical encryption module, which includes a hierarchical unit and an encryption unit; the hierarchical unit is used to classify the multi-source driving test data according to the privacy level of the data and determine the encryption level of the driving test data from different data sources; the encryption unit includes multiple encryption sub-units, each encryption sub-unit corresponding to an encryption level; the encryption sub-unit is used to encrypt the corresponding driving test data according to the encryption level and send the encrypted driving test data to the cloud-side device.

[0037] Specifically, a hierarchical encryption mechanism is established, dividing driver's license test data into four security levels: Level 1 data (core identity information such as user ID number and contact information), Level 2 data (sensitive business data such as exam scores and complete operation logs), Level 3 data (routine business data such as practice time and basic driving trajectory), and Level 4 data (non-sensitive data such as video template cache and temporary operation records). Differentiated encryption strategies are adopted for different data levels. Level 1 data uses a hybrid encryption algorithm of AES-256 and RSA, Level 2 data uses AES-128 encryption, Level 3 data uses SM4 encryption, and Level 4 data uses simple hash digest processing. In this embodiment, multiple encryption sub-units within the hierarchical and encryption units achieve privacy-first end-cloud collaboration, prioritizing privacy protection. Through collaborative work between the end-user device and the cloud server, user privacy is prioritized in all stages of data processing (collection, transmission, storage, and processing), while fully utilizing the computing resources of both the end-user device and the cloud to achieve efficient and secure generation of driver's license test videos.

[0038] In some embodiments of the present invention, the edge device stores a lightweight template library, and the cloud device stores a full-featured template library; the resource management module is specifically used to: determine the edge AI-driven video generation module as the target AI-driven video generation module when it is detected that the processing capability of the edge device meets the preset resource processing conditions and the lightweight template library contains a template library matching the video template type; and determine the cloud AI-driven video generation module as the target AI-driven video generation module when it is detected that the processing capability of the edge device does not meet the preset resource processing conditions, or the lightweight template library does not contain a template library matching the video template type.

[0039] Among these features, a comprehensive template library is built on cloud-side devices. A professional design team can create video templates with multiple (more than 20) subcategories tailored to the characteristics of driving test scenarios. These templates utilize patented data visualization graphic elements (such as score radar charts, trajectory heatmaps, and progress line animations) to form a template system with a unique visual style. Templates are stored in a layered architecture, consisting of a base layer (general layout, animation framework), a data layer (replaceable text and chart placeholders), and an effects layer (dynamic transitions and particle effects), facilitating flexible combination and rapid updates. The preset resource processing conditions can be that the CPU utilization rate of the terminal device is ≥30% and the GPU idle rate is ≤50%, and the network bandwidth is ≥5Mbps.

[0040] Specifically, by storing a lightweight template library on the edge device and a full-quantization template library on the cloud device, video generation processing is performed on the edge device when both template matching and computing power matching are met; if neither condition is met, video generation processing is performed on the cloud device, realizing dynamic collaboration between edge and cloud tasks. At the same time, privacy-first edge-cloud collaboration is achieved while ensuring privacy, thereby enabling efficient resource utilization and improving video generation efficiency.

[0041] In one specific implementation, a lightweight template library is built on the device side, containing various preset video templates, such as a "highlight celebration template" for excellent exam results, a "growth comparison template" for significant practice progress, and a "teaching demonstration template" for sharing driving test skills. The device-side AI-driven video generation module automatically selects the appropriate template based on the parsed data features using a decision tree matching strategy. For example, if the exam score is ≥90 points and the ranking is high, the "highlight celebration template" is selected; if the practice time for a certain item is reduced by more than 40% and the error rate decreases significantly, the "growth comparison template" is triggered. After selecting a template, the system automatically fills the parsed data into the corresponding animation, subtitle, chart, and other elements of the template, and calls the terminal GPU to complete the video synthesis locally. The entire process requires no manual editing by the user, achieving "zero-edit" video generation.

[0042] In another specific implementation, tasks are preferentially processed on the edge device when the following conditions are met: CPU utilization < 30% and GPU idle rate > 50%; a matching template exists in the local template library; and network bandwidth < 5Mbps or the network is unstable. In this case, tasks such as simple video compositing and basic data parsing are completed directly on the edge, reducing data uploads. When terminal resources are insufficient or complex processing is required (such as high-definition video rendering or large-scale template retrieval), processing is performed on the cloud-side device.

[0043] In some embodiments of the present invention, when the target AI-driven video generation module is an edge AI-driven video generation module, the edge AI-driven video generation module is used to: select a template matching the video template type from the lightweight template library to obtain a target video template, and fill the target video template with the driving test feature information corresponding to the driving test data to generate a target driving test video; when the target AI-driven video generation module is a cloud AI-driven video generation module, the cloud AI-driven video generation module is used to: select a template matching the video template type from the full-scale template library to obtain a target video template, fill the target video template with the driving test feature information corresponding to the driving test data to generate a target driving test video, and send the target driving test video to the edge device.

[0044] Specifically, when the resource management module determines that the "device-side priority" condition is met, the device-side AI-driven video generation module calls a template that matches the determined video template type from the lightweight template library to obtain the target video template. It then fills the driver's test feature information (scores, trajectory heatmaps, deduction tags, etc.) obtained after parsing the driver's test data into the template placeholders in order, and finally calls the terminal GPU to synthesize the target driver's test video with zero dependency. When the resource management module triggers the "cloud-side fallback" strategy, the cloud-side AI-driven video generation module is executed in the cloud-side device. Specifically, it selects a template that matches the video template type from the full-scale template library to obtain the target video template, fills the uploaded driver's test feature information after hierarchical encryption into the template, uses the GPU cluster to render in parallel to generate the target driver's test video, and sends the target driver's test video back to the device-side device.

[0045] In some embodiments of the present invention, the cloud-side device is further deployed with a security sandbox; the cloud-side AI-driven video generation module is further configured to: receive encrypted driving test data corresponding to driving test data transmitted to the security sandbox via HTTPS protocol; generate a target driving test video in the security sandbox based on the encrypted driving test data and the full-quantization template; encrypt the target driving test video using differential privacy technology and return it to the edge device.

[0046] Among them, a security sandbox is an isolated cloud environment that can create an independent sandbox environment on cloud-side devices through virtualization technology, isolating potentially untrusted or unknown software from the host system.

[0047] Specifically, the encrypted data is transmitted to a secure sandbox environment for processing. After processing, differential privacy technology is used to obfuscate the result, for example by adding controllable noise, before encrypting it again and returning it to the end device, ensuring the privacy and security of the data throughout the entire process.

[0048] In some embodiments of the present invention, the cloud-side device is further equipped with a blockchain evidence storage module; the cloud-side AI-driven video generation module is also used to upload the transmission and processing of the encrypted driver's test data to the blockchain evidence storage module.

[0049] Specifically, cloud-side devices use differential privacy technology to add controllable noise before encrypting and returning the data to the terminal. Throughout the process, data transmission records are synchronously uploaded to the blockchain's evidence storage module, leveraging the blockchain's immutability to record the data transmission and processing, ensuring data integrity and traceability.

[0050] In some embodiments of the present invention, the cloud-side device is further deployed with a template library update module, which includes a template scoring unit, a template visual element update unit, and a user-driven template update unit. The template scoring unit is used to periodically calculate the comprehensive score of each template in the full-quantification template library. The template visual element replacement unit is used to replace visual elements in templates with comprehensive scores less than a preset score threshold with target visual elements to generate subsequent templates. The target visual elements are visual elements extracted from templates of the same type with comprehensive scores greater than or equal to the preset score threshold using an AI element design model. The user-driven template update unit is used to obtain feedback data from target users regarding the use of the candidate templates within a preset number of periods, and select an updated template from the candidate templates based on the feedback data.

[0051] Specifically, the cloud-based dynamic template update mechanism keeps the full-format template library highly active, improves user satisfaction with the template library, and enables the system to quickly adapt to market changes and user needs, giving it strong scalability and competitiveness.

[0052] In one specific implementation, the template library update module collects template usage data from the previous 24 hours at 00:00 daily and calculates a comprehensive score for each template (share rate weighted at 40%, usage time weighted at 30%, and user rating weighted at 30%). For templates with scores below a threshold, the AI ​​design tool automatically extracts and replaces visual elements from similar high-scoring templates, generating 3-5 candidate versions. The system randomly selects 5% of active users for A / B testing, randomly assigning different template versions to users when using the video generation function. After 72 hours, based on user retention rate, number of secondary uses, rating feedback, and other data, the optimal version is selected to overwrite the original template, and an update notification is pushed to all user terminals.

[0053] In one specific implementation, such as Figure 2 The diagram shown is a flowchart of the template dynamic update process. Details are as follows: Data collection for template usage: The system first collects data generated when users use driving test video templates.

[0054] AI analytics uses data: It uses artificial intelligence technology to analyze collected data to evaluate the effectiveness of template usage and user satisfaction.

[0055] Template Score Below Threshold: After AI analysis, the system will determine if the template's score is below a preset threshold. AI Design Tool Optimizes the Template: If the template score is below the threshold, the AI ​​design tool will intervene to optimize the template. This includes adjusting the template's visual elements, animation rhythm, or data display format. Candidate Template Versions: Several candidate versions will be generated from the optimized template for further testing and evaluation.

[0056] A / B Testing Push to 5% of Users: These candidate template versions will be pushed to 5% of users through A / B testing to compare the effects of different versions.

[0057] Based on optimal user feedback, we collect user feedback on different template versions to determine which version is the most popular or performs best. Full release of the new template: If a candidate template version performs best in user feedback, that version will be selected as the new template and fully released to all users.

[0058] The template library on the device side will be updated synchronously with the new templates to ensure that all users can use the latest templates.

[0059] If no better template version is found after AI optimization and user testing, the system will continue to use the original template.

[0060] In some embodiments of the present invention, the edge AI-driven video generation module includes an edge AI parsing engine, a template decision unit, and a video synthesis unit; the edge AI parsing engine is used to extract driving test feature information from the driving test data; the template decision unit is used to parse the driving test feature information to determine a target video template; and the video synthesis unit is used to fill the driving test feature information into the target video template to generate a target driving test video.

[0061] The video synthesis unit can be a GPU unit on the edge device.

[0062] Specifically, the AI-driven video generation module on the device side analyzes the data in real time, triggers a template matching algorithm based on the analysis results, obtains a suitable template (i.e., the target video template) from the local machine or the cloud, automatically fills the data into the target video template, and calls the GPU of the device side to perform video synthesis, thus realizing the efficient generation of the target driving test video locally. In some embodiments of the present invention, the end-side device is further configured with a video preview unit and a video forwarding unit; the video preview unit is used to display the received target driving test video; the video forwarding unit is used to forward the received target driving test video to a social media platform.

[0063] Specifically, the generated target driving test videos can be previewed directly within the application and shared to social platforms such as WeChat and Douyin with one click. The zero-edit video generation mode driven by edge AI allows driving test users to complete video creation within 1 minute without professional skills. Compared with traditional manual editing, it improves video generation efficiency, greatly reduces the creation threshold, and stimulates users' enthusiasm for sharing.

[0064] In one specific implementation, such as Figure 3 The diagram shows the zero-edit video generation process, which involves: collecting data from the driving test process, including test scores, practice operation data, and vehicle driving trajectories; edge AI data analysis: the collected data is sent to the AI ​​analysis engine on the edge device for analysis and extraction of key information; template matching: the AI ​​analysis engine attempts to match a suitable video template in the local template library; if a suitable template is found, the process continues to the "automatic template filling" step; if not, the process redirects to "cloud template request"; cloud template request: if no suitable template is found on the edge, the system requests a template from the cloud server; downloading the template to the edge: a suitable template is downloaded from the cloud to the edge device; automatic template filling: whether the template is locally matched or downloaded from the cloud, the system automatically fills the template with driving test data; edge video compositing: the template with filled data is used for video compositing on the edge device to generate the final video file; video preview: users can preview the composited video to check their satisfaction; social media sharing: users can share the video on social media platforms to share their driving test experience with others.

[0065] In one specific implementation, such as Figure 4 The diagram shows the architecture of a privacy-first, edge-cloud collaborative system for generating driver's license test data with zero editing. The system consists of two main parts: edge devices and a cloud server, which interact through encrypted data and data characteristics.

[0066] The edge device includes a data acquisition unit responsible for collecting multi-source data from the driving test process, such as test scores, practice operation data, and vehicle driving trajectories; an edge AI parsing engine that uses a lightweight deep learning model to parse the collected data in real time and extract key information; a template matching decision tree that automatically selects appropriate video templates based on the parsed data features; a local video synthesizer that fills the parsed data into the positions of animation, subtitles, charts, and other elements corresponding to the template, and calls the terminal GPU to complete the video synthesis locally; and a lightweight template cache that stores preset video templates for quick matching and use.

[0067] The cloud server component includes: an encrypted storage database for storing encrypted user data to ensure data security; an AI computing cluster for handling complex computational tasks that edge devices cannot perform, such as high-definition video rendering and large-scale template retrieval; a template management system for managing the cloud-based video template library, including template design, storage, and updates; and a blockchain-based evidence storage system for recording data transmission and processing to ensure data integrity and immutability. Data acquisition: The edge device's data acquisition unit collects multi-source data during the driving test process; data encryption: The collected data is encrypted on the edge to ensure data security during transmission; data transmission: The encrypted data is transmitted to the encrypted storage database on the cloud server; data processing: The cloud server's AI computing cluster further processes the data, such as high-definition video rendering and large-scale template retrieval; template matching: The edge AI parsing engine automatically selects suitable templates based on the parsed data characteristics using a template matching decision tree; video compositing: The local video compositor fills the templates with data to generate video files; video sharing: The generated video files can be previewed directly within the application and shared to social media platforms with one click.

[0068] This system enables automated video creation of driver's license test data through the collaborative work of edge devices and cloud servers. Edge devices are responsible for data collection, encryption, and preliminary processing, while the cloud server handles data storage, complex calculations, and template management. A blockchain-based notarization system ensures data security and integrity, while a template management system enables dynamic updates and optimization of templates, providing driver's license test users with a secure and convenient video sharing solution.

[0069] like Figure 5 The diagram illustrates a privacy-first cloud-edge collaborative architecture, divided into three parts: data hierarchical encryption, task scheduling strategy, and secure transmission layer. Data acquisition: Driver's test data is first collected and undergoes hierarchical encryption. Task scheduling: Based on the resource status and network conditions of the terminal device, a decision is made whether to process the data on the terminal or in the cloud. Terminal processing: If the terminal processing conditions are met, the data is processed locally, such as video compositing and template filling. Cloud processing: If the terminal processing conditions are not met, the data is encrypted and transmitted to the cloud for processing, such as complex video rendering. Secure transmission: All transmitted data is protected using HTTPS protocol and blockchain hash notarization technology to ensure data security. This design approach comprehensively considers data security, processing efficiency, and user experience in the driver's test video generation system. Through hierarchical encryption and intelligent task scheduling, the system can efficiently utilize computing resources and improve video generation efficiency while protecting user privacy.

[0070] like Figure 6 As shown, the present invention also provides a method for generating driver's test videos, which is applied to the driver's test video generation system in the above embodiments. Figure 6A schematic flowchart of an embodiment of the driver's test video generation method provided by the present invention includes: S601, acquires driving test data from multiple data sources during the driving test process; S602, parse the driving test data and determine the corresponding video template type; S603, based on the video template type and the processing capability of the terminal device, determine the target AI-driven video generation module, wherein the target AI-driven video generation module is either a terminal AI-driven video generation module or a cloud AI-driven video generation module; S604, Based on the driving test data and the video template type, generate the target driving test video through the target AI-driven video generation module.

[0071] It should be noted that the segment-side device and cloud-side device in the driver's test video generation system provided in the above embodiments can implement the technical solutions described in the above driver's test video generation method embodiments, and will not be repeated here.

[0072] like Figure 7 The diagram shown illustrates the timing of the edge-cloud task scheduling, illustrating the interaction between the edge device, the cloud server, and the data encryption layer. The edge device is responsible for data acquisition and initial processing, the cloud server handles high-load tasks, and the data encryption layer ensures data security and privacy protection. Through this collaborative approach, the system can efficiently generate driver's license test videos while protecting user privacy.

[0073] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A driving test video generation system, characterized in that, include: The driver's test data acquisition module, the edge AI-driven video generation module, and the resource management module are deployed on the edge device, and the cloud AI-driven video generation module is deployed on the cloud device. The processing power of the cloud AI-driven video generation module is higher than that of the edge AI-driven video generation module. The driving test data acquisition module is used to collect driving test data from multiple data sources during the driving test process and send the driving test data to the AI-driven video generation module on the edge. The edge AI-driven video generation module is used to parse the driving test data and determine the corresponding video template type; The resource management module is used to determine the target AI-driven video generation module as either the edge AI-driven video generation module or the cloud AI-driven video generation module based on the video template type and the processing capability of the edge device. The target AI-driven video generation module is used to generate target driving test videos based on the driving test data and the video template type.

2. The driver's license test video generation system according to claim 1, characterized in that, The end-side device is also equipped with a hierarchical encryption module, which includes a hierarchical unit and an encryption unit. The grading unit is used to grade the multi-source driver's test data according to the privacy level of the data, and to determine the encryption level of driver's test data from different data sources. The encryption unit includes multiple encryption sub-units, each of which corresponds to an encryption level. The encryption sub-unit is used to encrypt the corresponding driving test data according to the encryption level and send the encrypted driving test data to the cloud-side device.

3. The driver's license test video generation system according to claim 2, characterized in that, The edge device stores a lightweight template library, and the cloud device stores a full-featured template library; the resource management module is specifically used for: If the processing capability of the edge device meets the preset resource processing conditions and the lightweight template library contains a template library that matches the video template type, the edge AI-driven video generation module is determined to be the target AI-driven video generation module. If the processing capability of the edge device is not met by the preset resource processing conditions, or if the lightweight template library does not have a template library that matches the video template type, the cloud-side AI-driven video generation module is determined to be the target AI-driven video generation module.

4. The driver's license test video generation system according to claim 3, characterized in that, When the target AI-driven video generation module is an edge-side AI-driven video generation module, the edge-side AI-driven video generation module is used for: A template matching the video template type is selected from the lightweight template library to obtain a target video template, and the driving test feature information corresponding to the driving test data is filled into the target video template to generate a target driving test video; When the target AI-driven video generation module is a cloud-based AI-driven video generation module, the cloud-based AI-driven video generation module is used for: After selecting a template that matches the video template type from the full-quantization template library to obtain the target video template, the driving test feature information corresponding to the driving test data is filled into the target video template to generate the target driving test video, and the target driving test video is sent to the end device.

5. The driver's license test video generation system according to claim 4, characterized in that, The cloud-side device is also equipped with a security sandbox; the cloud-side AI-driven video generation module is also used for: Receive encrypted driving test data corresponding to the driving test data transmitted to the security sandbox via HTTPS protocol; Based on the encrypted driving test data and the fully quantized template, a target driving test video is generated in the secure sandbox. The target driving test video is blurred and encrypted using differential privacy technology, and then returned to the end device.

6. The driver's license test video generation system according to claim 5, characterized in that, The cloud-side device is also equipped with a blockchain evidence storage module; the cloud-side AI-driven video generation module is also used for: The transmission and processing of the encrypted driver's test data are uploaded to the blockchain evidence storage module.

7. The driver's license test video generation system according to claim 3, characterized in that, The cloud-side device is also equipped with a template library update module, which includes a template scoring unit, a template visual element update unit, and a user-driven template update unit. The template scoring unit is used to periodically calculate the comprehensive score of each template in the full-quantification template library; The template visual element replacement unit is used to replace visual elements in templates with a comprehensive score less than a preset score threshold with target visual elements to generate subsequent templates. The target visual elements are visual elements extracted from templates of the same type with a comprehensive score greater than or equal to the preset score threshold by an AI element design model. The user-driven template update unit is used to obtain feedback data from target users regarding the use of the candidate templates within a preset number of periods, and to select an update template from the candidate templates based on the feedback data.

8. The driver's license test video generation system according to claim 4, characterized in that, The edge AI-driven video generation module includes an edge AI parsing engine, a template decision unit, and a video synthesis unit; The edge AI analysis engine is used to extract driving test feature information from the driving test data; The template decision unit is used to analyze the driving test feature information and determine the target video template; The video synthesis unit is used to fill the driving test feature information into the target video template to generate the target driving test video.

9. The driver's license test video generation system according to claim 1, characterized in that, The end-side device is also equipped with a video preview unit and a video forwarding unit; The video preview unit is used to display the received target driving test video; The video forwarding unit is used to forward the received target driving test video to a social media platform.

10. A method for generating driver's license test videos, characterized in that, The method, applied to the driver's test video generation system according to any one of claims 1-9, comprises: Acquire driving test data from multiple data sources during the driving test process; The driving test data is analyzed to determine the corresponding video template type; Based on the video template type and the processing capability of the edge device, a target AI-driven video generation module is determined, wherein the target AI-driven video generation module is either an edge AI-driven video generation module or a cloud AI-driven video generation module. Based on the driving test data and the video template type, the target driving test video is generated by the target AI-driven video generation module.

Images