Country short video automatic generation platform based on AI drive

Abstract

The invention relates to the technical field of automatic generation of short videos, in particular to an AI-driven automatic generation platform for rural short videos, which comprises a system architecture module. An overall architecture module, a core module design module, a team division module, a cost reduction design module and a list confirmation module are arranged in the system architecture module; and a user side module and a technology stack type selection module are arranged in the overall architecture module. According to the invention, based on the rural ecology, a corresponding AI-driven short video generation platform can be established according to the rural ecological condition, the corresponding rural propaganda video of the platform is generated, the video can be ensured to fully fit the rural ecological environment, and the characteristics of the rural can be effectively highlighted; therefore, a virtuous circle of'technology enabling-value creation-ecological feedback 'is finally formed, and the method becomes a benchmark infrastructure for digital reform.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic short video generation, and in particular to an AI-driven automatic generation platform for rural short videos. Background Art

[0002] Rural-themed short videos have become an important vehicle for disseminating local culture and promoting industrial upgrading. However, issues such as content homogeneity, high barriers to entry, and difficulty in monetizing traffic have long constrained their development.

[0003] Local governments are increasingly demanding digital content dissemination for rural areas, particularly in the areas of cultural tourism promotion and agricultural product marketing. For example, the application of digital communication technologies and the rise of new formats such as e-commerce for agricultural products and online agricultural assistance demonstrate the crucial role of digitalization. Furthermore, according to the "2023 China Rural Digital Development Research Report," rural digitalization has achieved significant success in parts of the country, further demonstrating the strong demand for digital content dissemination from local governments and providing opportunities for platforms to collaborate with relevant departments to secure policy support and market procurement orders.

[0004] The AI-driven automatic short video generation system, with publication number CN118413711A, includes an acquisition module for acquiring client demand information and matching the demand information with a target video; an extraction module connected to the acquisition module for acquiring a target video and extracting video information from the target video to obtain feature information; a classification module connected to the extraction module for classifying the feature information to obtain multiple sub-feature information; an editing module connected to the classification module for matching corresponding editing information in a database based on the demand information and the feature information; and a processing module connected to the editing module for processing the target video based on the editing information to obtain a pre-processed short video. The present invention can ensure the stability of the special effects applied to the video, ensure the stability of the picture, have a good short video synthesis effect, and can quickly match the corresponding processing method information in subsequent matching.

[0005] How to prepare short videos that are more suitable for rural conditions is a difficult problem in this field. Summary of the Invention

[0006] The purpose of this invention is to solve the shortcomings of the existing technology and propose an AI-driven automatic generation platform for rural short videos.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] The AI-driven automatic generation platform for rural short videos includes a system architecture module; the system architecture module includes an overall architecture module, a core module design module, a team division of labor module, a cost reduction design module, and a list confirmation module;

[0009] The overall architecture module is provided with a user terminal module and a technology stack selection module;

[0010] The core module design module includes an AI module generation module, a dialect language and subtitle generation module, a traffic sharing alliance module, and a data service and analysis module;

[0011] The team division of labor module is provided with a front-end group module, a back-end group module, an AI group module and an operation and maintenance group module.

[0012] Compared with the existing technology, this application can be based on rural ecology, so that a corresponding AI-driven short video generation platform can be established according to the rural ecological conditions, and by generating corresponding rural promotional videos on the platform, it can ensure that the video fully fits the rural ecological environment and can effectively highlight the characteristics of the countryside, so as to ultimately form a virtuous cycle of "technology empowerment-value creation-ecological feedback" and become a benchmark infrastructure for digital reform.

[0013] Preferably, the user-end module is provided with a front-end server module, a back-end service cluster module, an AI processing layer module, a data layer and a third-party interface module;

[0014] The technology stack selection module includes a front-end module, a database module and a deployment module.

[0015] Furthermore, the front-end server is an API gateway; the back-end service cluster is a microservice architecture; the AI processing layer can implement template generation, voice recognition, and subtitle synthesis; the data layer can set the material library, user data, and traffic distribution data; third-party interfaces include Douyin / Kuaishou API, iFlytek voice recognition, and blockchain evidence storage; front-end: uses React (Web), UniApp (App), and Vue.js (interface interaction); back-end: uses Python (Flask / Django) and Node.js (high concurrency scenarios); AI processing: Python (TensorFlow / PyTorch), FFmpeg (video processing), iFlytek open platform (dialect recognition); database: MongoDB (unstructured data), MySQL (structured data), Redis (cache); deployment: Docker containerization, Kubernetes cluster, CDN acceleration (video distribution);

[0016] The above technical solution can establish a relatively complete technical architecture for the platform, making it easy for users to import materials and automatically generate short videos. At the same time, it can also quickly realize video uploading and subsequent video traffic promotion, monitoring and other functions.

[0017] Preferably, the AI module generation module is provided with a material parsing module, a module matching module, a content filling module and an output video module.

[0018] Furthermore: the function of the AI template generation module is to automatically generate short videos that conform to the rural theme based on the materials (pictures, video clips, text) input by the user;

[0019] Its operation logic is as follows:

[0020] Step 1: Material analysis: Extract visual features (such as scenes, objects) and text keywords (such as "farmland" and "harvest") from images / videos;

[0021] Step 2: Template matching: Match the optimal template from the preset template library (such as "agricultural explanation" and "cultural tourism promotion");

[0022] Step 3: Content filling:

[0023] Automatic editing: splicing materials according to the template rhythm (e.g. 5 seconds of farmland panorama at the beginning, 10 seconds of character interview in the middle, and 5 seconds of product close-up at the end);

[0024] Music adaptation: Match background music according to the scene (such as folk music, natural sound effects);

[0025] Step 4: Output video: adaptive resolution (supports vertical / horizontal screen) and generates video files in multiple formats;

[0026] Technical details to note:

[0027] The template library is stored in JSON format, including storyboards, music paths, and special effect parameters;

[0028] Use OpenCV for video editing and FFmpeg for transcoding and compression.

[0029] Preferably, the dialect speech and subtitle generation module is provided with a speech recognition module, a multilingual translation module and a dynamic subtitle synthesis module.

[0030] Furthermore: the dialect speech recognition and subtitle generation functions support speech recognition of dialects such as Minnan and Cantonese, and generate multi-language subtitles in real time;

[0031] Its operation logic is as follows:

[0032] Step 1: Voice Recognition:

[0033] Call the iFLYTEK API and pass in the audio file (noise reduction preprocessing is required).

[0034] Returns the text and language tag (e.g. "Cantonese").

[0035] Step 2: Multilingual Translation:

[0036] If the user selects "English subtitles", the Google Translate API is called to translate the text.

[0037] Step 3: Dynamic subtitle synthesis:

[0038] Use the pysrt library to generate SRT subtitle files, and the timeline is synchronized with the voice.

[0039] Subtitle style: Chinese brush font, position adaptive (to avoid blocking people's faces).

[0040] Its optimization points:

[0041] Cache high-frequency vocabulary (such as agricultural terms such as "fertilize" and "harvest") to reduce API call latency.

[0042] Support offline recognition (lightweight models such as MozillaDeepSpeech).

[0043] Preferably, the traffic sharing alliance module is provided with a traffic pool construction module, an exposure algorithm module, an advertising profit sharing module and a technical tool module.

[0044] Furthermore: the function of the traffic sharing alliance is to integrate the traffic of villagers, self-media, and cultural and tourism bureaus, and optimize exposure distribution through algorithms;

[0045] Its implementation logic:

[0046] Step 1: Traffic pool construction:

[0047] Self-media entry: provide RSS interface to synchronize Tik Tok / Kuaishou videos;

[0048] Government cooperation: access to the content library of the county-level integrated media center;

[0049] Step 2: Exposure algorithm:

[0050] Allocate traffic based on user interests (such as the frequency of watching videos on agriculture, rural areas and farmers) and LBS (geographic location);

[0051] Top accounts (such as county-level influencers) are given priority in content recommendation, driving exposure for grassroots accounts;

[0052] Step 3: Advertising profit sharing:

[0053] Insert brand advertisements (such as agricultural product logos) into videos, and the platform takes a 10%-20% commission;

[0054] Public service advertisements (such as anti-drug propaganda) are purchased by the government at its expense;

[0055] Step 4: Technical Tools:

[0056] RabbitMQ is used to implement traffic distribution queues; the recommendation algorithm is based on collaborative filtering (user behavior similarity).

[0057] Preferably, the data service and analysis module is provided with a data acquisition module and an analysis report module.

[0058] Further: Data services and analysis functions: providing the government with communication effect reports (such as video clicks and geographical distribution).

[0059] Data flow:

[0060] Step 1: Data Collection:

[0061] Tracking statistics: user playback time, like / forward behavior, and device information.

[0062] Log storage: Elasticsearch (real-time query), Hadoop (batch analysis).

[0063] Step 2: Analysis Report:

[0064] Generate visual reports (Echarts / Tableau), such as "The dissemination effect of a certain type of video in western Guangdong."

[0065] Supports API data export (e.g. `api / report / tourism` returns cultural tourism video data).

[0066] Preferably, the material parsing module has the function of extracting visual features and text keywords of pictures / videos;

[0067] The template matching module has the function of matching the optimal template from the preset template library

[0068] The content filling module is provided with an automatic editing module and a music adaptation module;

[0069] The automatic editing module has the function of splicing materials according to the template rhythm;

[0070] The music adaptation module has the function of matching background music according to the scene

[0071] The output video module has the functions of resolution adaptation and generation of multi-format files.

[0072] Preferably, the speech recognition module has the function of calling the iFLYTEK API, inputting the noise-reduced audio file, and returning the text with language tags;

[0073] The multilingual translation module has the function of translating text through Google Translate API according to user selection and producing corresponding subtitles;

[0074] The dynamic subtitle synthesis module generates an SRT subtitle file using a font library and ensures that the timeline is synchronized with the voice.

[0075] Preferably, the traffic pool construction module includes a self-media entry module and a government cooperation module; the self-media entry module can provide an interface for synchronization with the short video platform; the government cooperation module can access the content library of the county-level integrated media center;

[0076] The exposure algorithm module can allocate traffic based on user interests and geographic location; and can give priority to recommending top accounts that have driven linked accounts;

[0077] The advertising profit sharing module is used to process the costs of product advertising and public service advertising;

[0078] The technical tool module uses RabbitMQ to implement traffic distribution queues.

[0079] Preferably, the data acquisition module is provided with a point-of-sale statistics module and a log storage module;

[0080] The tracking statistics module is used to count the user's playback time, like / forward behavior, and device information; the log storage module is equipped with a real-time query module and a batch analysis module;

[0081] The analysis report module can generate visual reports and support API data export function. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 This is a structural diagram of the AI-driven rural short video automatic generation platform proposed in this invention;

[0083] Figure 2 This is a structural block diagram of the AI module generation module of the AI-driven rural short video automatic generation platform proposed in the present invention;

[0084] Figure 3 This is a structural diagram of the dialect language and subtitle generation module of the AI-driven rural short video automatic generation platform proposed by the present invention;

[0085] Figure 4 This is a structural diagram of the medium-traffic sharing alliance based on the AI-driven rural short video automatic generation platform proposed in the present invention;

[0086] Figure 5 This is a structural diagram of the data service and analysis of the AI-driven rural short video automatic generation platform proposed in the present invention. DETAILED DESCRIPTION

[0087] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0088] Reference Figure 1-5 , an AI-driven automatic generation platform for rural short videos, including a system architecture module; the system architecture module is equipped with an overall architecture module, a core module design module, a team division of labor module, and a cost reduction design module to fully reduce the required costs. For the low accuracy of dialect recognition, it is possible to cooperate with universities to annotate the dialect corpus and optimize the ASR model; for high server costs, consider using Alibaba Cloud burst performance instances, which automatically shrink at night. For the most important content copyright disputes, a material authorization agreement will be signed with the photographer / villager and infringement insurance will be purchased; a list confirmation module; the operation process of the list confirmation module is as follows: Step 1, code repository: GitHub hosting, with detailed comments and API documentation; Step 2, deployment documents: Docker Compose files, Kubernetes configuration files; Step 3, test report: covering AI template generation, dialect subtitles, and high-concurrency stress testing; Step 4, user manual: a "three-step operation guide" graphic manual for villagers.

[0089] The overall architecture module includes a user-end module and a technology stack selection module.

[0090] The core module design module includes AI module generation module, dialect language and subtitle generation module, traffic sharing alliance module, and data service and analysis module;

[0091] The team division of labor module includes front-end group module, back-end group module, AI group module and operation and maintenance group module; the team division of labor module is set up as follows: front-end group: responsible for user interface, video preview, and multi-platform adaptation; back-end group: development of API gateway, traffic scheduling algorithm, and data interface; AI group: optimization of template generation model, training of dialect recognition lightweight model; operation and maintenance group: building Docker containers, monitoring server load (Prometheus+Grafana).

[0092] In the present invention, the user-side module is equipped with a front-end server module, a back-end service cluster module, an AI processing layer module, a data layer and a third-party interface module; the technology stack selection module includes a front-end module, a database module and a deployment module; the front-end server is an API gateway; the back-end service cluster is a microservice architecture; the AI processing layer can realize template generation, speech recognition and subtitle synthesis; the data layer can set the material library, user data and traffic distribution data; the third-party interface includes Douyin / Kuaishou API, iFlytek speech recognition and blockchain evidence storage; the front-end: uses React (Web), UniApp (App) and Vue.js (interface interaction); the back-end: uses Python (Flask / Django) and Node.js (high concurrency scenarios); AI processing: Python (TensorFlow / PyTorch), FFmpeg (video processing) and iFlytek open platform (dialect recognition); database: MongoDB (unstructured data), MySQL (structured data) and Redis (cache); deployment: Docker containerization, Kubernetes cluster and CDN acceleration (video distribution);

[0093] The above technical solution can establish a relatively complete technical architecture for the platform, making it easy for users to import materials and automatically generate short videos. At the same time, it can also quickly realize video uploading and subsequent video traffic promotion, monitoring and other functions.

[0094] In the present invention, the AI module generation module is provided with a material analysis module, a module matching module, a content filling module and a video output module; the material analysis module has the function of extracting the visual features and text keywords of the picture / video;

[0095] The template matching module has the function of matching the optimal template from the preset template library

[0096] The content filling module is equipped with an automatic editing module and a music adaptation module;

[0097] The automatic editing module has the function of splicing materials according to the rhythm of the template;

[0098] The music adaptation module has the function of matching background music according to the scene

[0099] The output video module has the functions of resolution adaptation and generating multi-format files.

[0100] In the present invention, the dialect speech and subtitle generation module is equipped with a speech recognition module, a multilingual translation module and a dynamic subtitle synthesis module; the speech recognition module has the function of calling the iFlytek API, inputting the noise-reduced audio file, and returning the text to add language tags;

[0101] The multilingual translation module has the function of translating texts through Google Translate API and producing corresponding subtitles according to user selections;

[0102] The dynamic subtitle synthesis module uses the font library to generate SRT subtitle files and ensures that the timeline is synchronized with the voice.

[0103] In the present invention, the traffic sharing alliance module is equipped with a traffic pool construction module, an exposure algorithm module, an advertising profit sharing module, and a technical tool module; the traffic pool construction module includes a self-media entry module and a government cooperation module; the self-media entry module can provide an interface for synchronizing short video platforms; the government cooperation module can access the content library of the county-level integrated media center;

[0104] The exposure algorithm module can allocate traffic based on user interests and geographic location; and can prioritize top accounts and their associated accounts;

[0105] The advertising profit sharing module is used to process the costs of product advertising and public service advertising;

[0106] The technical tool module uses RabbitMQ to implement traffic distribution queues.

[0107] In the present invention, the data service and analysis module is provided with a data acquisition module and an analysis report module; the data acquisition module is provided with a buried point statistics module and a log storage module;

[0108] The tracking statistics module is used to count user playback time, like / forward behavior, and device information; the log storage module is equipped with a real-time query module and a batch analysis module;

[0109] The analysis report module can generate visual reports and support API data export function.

[0110] In the present invention, in the actual application process, a cold start strategy can be adopted, that is, first by selecting a corresponding number of pilot villages for cooperation, the pilot villages have promotion advantages: tourism or agriculture; provide this platform free of charge and conduct training to quickly generate cases; at the same time, cooperate with the county-level integrated media center where the pilot villages are located to obtain initial promotion and cooperation;

[0111] At the same time, this platform can collect rural pictures and edit rural texts to achieve rapid generation of rural videos, and can set up AI virtual anchors according to the situation to help reduce risks.

[0112] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An AI-driven automatic rural short video generation platform, including a system architecture module; the system architecture module includes an overall architecture module, a core module design module, a team division of labor module, a cost reduction design module, and a checklist confirmation module; The overall architecture module is provided with a user terminal module and a technology stack selection module; The core module design module includes an AI module generation module, a dialect language and subtitle generation module, a traffic sharing alliance module, and a data service and analysis module; The team division of labor module is provided with a front-end group module, a back-end group module, an AI group module and an operation and maintenance group module.

2. The AI-driven rural short video automatic generation platform according to claim 1 is characterized by: The user-side module is equipped with a front-end server module, a back-end service cluster module, an AI processing layer module, a data layer and a third-party interface module; The technology stack selection module includes a front-end module, a database module and a deployment module.

3. The AI-driven rural short video automatic generation platform according to claim 1 is characterized by: The AI module generation module is provided with a material parsing module, a module matching module, a content filling module and an output video module.

4. The AI-driven rural short video automatic generation platform according to claim 1 is characterized by: The dialect speech and subtitle generation module is provided with a speech recognition module, a multilingual translation module and a dynamic subtitle synthesis module.

5. The AI-driven rural short video automatic generation platform according to claim 1 is characterized by: The traffic sharing alliance module is equipped with a traffic pool construction module, an exposure algorithm module, an advertising profit sharing module and a technical tool module.

6. The AI-driven rural short video automatic generation platform according to claim 1 is characterized by: The data service and analysis module is provided with a data acquisition module and an analysis report module.

7. The AI-driven rural short video automatic generation platform according to claim 3 is characterized by: The material analysis module has the function of extracting visual features and text keywords of pictures / videos; The template matching module has the function of matching the optimal template from the preset template library The content filling module is provided with an automatic editing module and a music adaptation module; The automatic editing module has the function of splicing materials according to the template rhythm; The music adaptation module has the function of matching background music according to the scene The output video module has the functions of resolution adaptation and generation of multi-format files.

8. The AI-driven rural short video automatic generation platform according to claim 4 is characterized by: The speech recognition module has the function of calling the iFlytek API, inputting the noise-reduced audio file, and returning the text with language tags; The multilingual translation module has the function of translating text through Google Translate API according to user selection and producing corresponding subtitles; The dynamic subtitle synthesis module generates an SRT subtitle file using a font library and ensures that the timeline is synchronized with the voice.

9. The AI-driven rural short video automatic generation platform according to claim 5 is characterized by: The traffic pool construction module includes a self-media entry module and a government cooperation module; the self-media entry module can provide an interface for synchronization with the short video platform; the government cooperation module can access the content library of the county-level integrated media center; The exposure algorithm module can allocate traffic based on user interests and geographic location; and can give priority to recommending top accounts that have driven linked accounts; The advertising profit sharing module is used to process the costs of product advertising and public service advertising; The technical tool module uses RabbitMQ to implement traffic distribution queues.

10. The AI-driven rural short video automatic generation platform according to claim 6 is characterized by: The data acquisition module is provided with a buried point statistics module and a log storage module; The tracking statistics module is used to count the user's playback time, like / forward behavior, and device information; the log storage module is equipped with a real-time query module and a batch analysis module; The analysis report module can generate visual reports and support API data export function.

Images