AI material generation acceleration method for e-commerce publicity

Abstract

The invention discloses an AI material generation acceleration method for e-commerce propaganda, and relates to the technical field of e-commerce propaganda, and the method comprises the following steps: at the beginning stage of video editing, firstly processing a video to be edited through a preset editing period by an AI technology, and guaranteeing that the initial stage of video editing reaches a quality standard; according to the method, the video complexity is intelligently evaluated, the editing time is dynamically adjusted, the editing period is shortened, the manufacturing process is accelerated, the production efficiency is improved, it is ensured that an e-commerce platform responds to market activities in time under the high-concurrency requirement, and key promotion opportunities are prevented from being missed. And for a video with a complex concept, a preset editing period is continuously adopted, and deep analysis of emotion fluctuation, visual contrast and other characteristics is combined with a long and short-term memory network, so that video quality maximization is ensured, logicality, fluency and visual effect are accurately adjusted and optimized, and user experience, brand image and conversion rate are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce promotion, and in particular to an AI material generation acceleration method for e-commerce promotion. Background Art

[0002] AI material generation acceleration for e-commerce promotion is to improve the generation efficiency and quality of marketing materials (including pictures, videos, and texts) through artificial intelligence technology, and provide efficient promotion support for e-commerce platforms. This method combines deep learning models with high-quality industry data sets to build an adaptable generation framework to achieve accurate understanding and expression of product characteristics and promotion scenarios. At the same time, a reinforcement learning mechanism is introduced to dynamically optimize the generation process, thereby enhancing the attractiveness and personalized effect of marketing materials. The method covers key links such as data preprocessing, generation model training, automatic generation, effect evaluation, and iterative optimization to ensure that the generated materials meet the marketing needs of e-commerce and have high visual impact and copywriting creativity. Through this method, e-commerce companies can quickly generate diversified and high-quality marketing materials, reduce labor input costs, significantly improve the response speed and promotion effect of promotional activities, and gain advantages in the fiercely competitive market.

[0003] In e-commerce promotion, the role of video editing through AI is mainly to improve the efficiency, accuracy and personalization of video production, so that e-commerce platforms can quickly respond to market demand and effectively attract consumers. AI technology can automatically analyze key elements in the video, such as product close-ups, usage scenarios, action changes, etc., identify and retain the most attractive shots, remove redundant content, and optimize the video structure. This intelligent editing not only enhances the visual impact of the video, but also automatically adjusts the sequence, duration and transition effects of the shots according to different marketing goals and audience needs to ensure that the video content is more attractive and appealing. In addition, AI editing can also optimize video performance in combination with data feedback, and improve the audience's viewing experience and conversion rate by continuously iterating and optimizing content. Through automated video editing, e-commerce platforms can generate high-quality marketing videos in a short period of time, significantly reduce labor costs, and improve the efficiency and effectiveness of promotional activities.

[0004] The prior art has the following deficiencies: When using AI to edit videos, the existing technology usually sets a longer editing time to ensure the maximum video quality. This is because video editing is not just about removing unnecessary parts, but also requires fine optimization and adjustment of the logic, fluency, visual impact and target audience needs of the content. However, when the video content to be edited is relatively simple, continuing to use the set longer editing time may lead to serious waste of efficiency and cost. Although AI can ensure video quality, if the editing time is too long, the computing resources and time consumed by it do not match the complexity of the video content, resulting in resource waste. This not only increases the hardware burden, but also slows down the overall video production progress, especially when e-commerce platforms need high concurrency (such as during promotional activities such as Double Eleven and Black Friday) to generate promotional videos, they may not be able to respond to market demand in time and miss the best marketing opportunities. Ultimately, this mismatched editing time will affect project efficiency, delay the launch of marketing activities, and thus reduce the market competitiveness of the platform.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide an AI material generation acceleration method for e-commerce promotion, which can intelligently evaluate the complexity of the video and dynamically adjust the editing time, especially for "simple-concept" videos, shorten the editing cycle to accelerate the production process, improve production efficiency, and ensure that the e-commerce platform can respond to market activities in a timely manner under high concurrency requirements to avoid missing key promotion opportunities. For "complex-concept" videos, the preset editing cycle is continued to be used, combined with the deep analysis of features such as emotional fluctuations and visual contrast by the long short-term memory network (LSTM), to ensure the maximum video quality, accurately adjust the logic, fluency and visual effects, thereby improving the user experience, brand image and conversion rate, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above objectives, the present invention provides the following technical solutions: 1. An AI material generation acceleration method for e-commerce promotion, comprising the following steps: At the beginning of video editing, AI technology first processes the video to be edited through a pre-set editing cycle to ensure that the initial stage of video editing meets the quality standards; During the video editing process, AI will monitor and obtain various parameter information of the video content in real time, dynamically capture the real-time data of the video content, and provide a basis for complexity assessment; Preprocess the acquired data information, purify and normalize the data to facilitate subsequent feature extraction and analysis; Extract key indicators reflecting video complexity from preprocessed data, analyze the extracted indicators in detail within the detection window, and comprehensively and dynamically quantify the video complexity through the analyzed features; The quantified complexity index is input into the pre-trained long short-term memory network, and the video complexity is intelligently evaluated through the long short-term memory network to ensure the accuracy and reliability of the evaluation results; Based on the evaluation results of the long short-term memory network, the video content is divided into two categories: "simple conception" and "complex conception"; For videos classified as "complex in conception", the preset editing cycle will continue to be used to ensure maximum video quality and to conduct more detailed optimization and adjustments; For videos classified as "simple in conception", the editing time is dynamically shortened based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining editing quality.

[0008] Preferably, the specific steps of AI technology processing the video to be edited through a preset editing cycle are as follows: In the initial stage of video editing, AI technology is used to analyze the overall content of the video to be edited; Based on the preset editing cycle, AI technology will apply standardized editing rules to process video content; Based on the application of editing rules, AI technology begins to automatically edit videos.

[0009] Preferably, key indicators reflecting the complexity of the video are extracted from the preprocessed data, and the extracted features include the degree of fluctuation of emotional expression in the video and the contrast between picture elements in the video. Under the detection window, the extracted degree of fluctuation of emotional expression in the video and the contrast between picture elements in the video are analyzed to generate emotional fluctuation reference values ​​and visual contrast reference values, respectively, and the complexity of the video is quantified by the emotional fluctuation reference values ​​and the visual contrast reference values.

[0010] Preferably, the specific steps of analyzing the fluctuation degree of emotional expression in the video under the detection window to generate the emotional fluctuation reference value are as follows: First, the speech and visual information in the video is processed through the emotion recognition algorithm to extract the emotional features of each frame in the video. For each frame, the facial expressions, voice intonation and visual content in the video are analyzed by multimodal emotion through AI technology to construct the emotion feature vector. The construction formula is as follows: , where is a visual emotion feature, indicating Emotional features extracted from video content in frames, is the audio emotion feature, indicating The emotional features obtained through speech analysis in the frame, is the text sentiment feature, indicating The emotional features extracted from the frame by text analysis, , as well as are weight coefficients, is the sentiment feature vector, indicating The comprehensive reflection of all emotional features in the frame video; After extracting the emotional features of each frame, the changing trend of the emotion is analyzed, and the local difference method is used to quantify the amplitude of the emotional change. and , calculate the change range of emotional features, the calculation expression is as follows: , where is the emotional feature vector of the next frame, is the fluctuation range control factor, It is the first Frame and The magnitude of emotional change between frames.

[0011] Preferably, after calculating the amplitude of the emotional fluctuation of each pair of adjacent frames, the frequency and intensity of the overall emotional fluctuation are evaluated, and the evaluation formula is as follows: , where It is the reference value of emotional fluctuation. It is the first Frame and The magnitude of emotional change between frames, is the time decay factor, is the attenuation coefficient, is the total number of frames in the video; The extracted emotional fluctuation information is integrated through nonlinear functions to generate emotional fluctuation reference values. The generation formula is as follows: , where It is the reference value of emotional fluctuation. is a nonlinear adjustment parameter.

[0012] Preferably, the specific steps of analyzing the contrast between the picture elements in the video under the detection window to generate the visual contrast reference value are as follows: First, the video frame is preprocessed to extract the key information in the video. After preprocessing, the brightness and chromaticity information of each pixel is extracted, and the contrast features of each pixel area are extracted using local contrast. The formula is as follows: , where Indicates that in the video frame, the position The local contrast of the pixel at is the pixel coordinate, It's a pixel The brightness value, and is the offset of the convolution window, which is used to specify the position of adjacent pixels, is the radius of the convolution window; Next, by calculating the contrast difference between adjacent areas in the video frame, the visual impact of the overall video is quantified. In order to accurately capture the contrast changes between the elements of the picture, the contrast difference factor is introduced. The calculation formula of the contrast difference factor is as follows: , where is the total number of pixels, It's a pixel The local contrast value, is the contrast difference factor.

[0013] Preferably, the spatial distribution in the video frame is comprehensively analyzed, and the spatial contrast factor is used to comprehensively evaluate the spatial contrast of the entire frame. The evaluation formula is as follows: , where is the spatial contrast factor, is the spatial weight factor; Comprehensive pixel local contrast , contrast difference factor and the spatial contrast factor Generate visual contrast reference value, the generation formula is as follows: , where is the visual contrast reference value, is the contrast difference factor The weight coefficient of is the inter-contrast factor The weight coefficient of is the global brightness contrast adjustment factor.

[0014] Preferably, the emotion fluctuation reference value and the visual contrast reference value generated after analysis are input into a pre-trained long short-term memory network, a video complexity index is generated through the long short-term memory network, and the video complexity is intelligently evaluated through the video complexity index.

[0015] Preferably, the video complexity index generated when the video complexity is intelligently evaluated by the pre-trained long short-term memory network is compared and analyzed with a pre-set video complexity index reference threshold, and the video content is divided, and the division steps are as follows: If the video complexity index is greater than or equal to a preset video complexity index reference threshold, the current video content is classified as complex in conception; If the video complexity index is less than a preset video complexity index reference threshold, the current video content is classified as having a simple concept.

[0016] Preferably, for videos classified as "simple in conception", the specific steps of dynamically shortening the editing time based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining the editing quality are as follows: After the video is judged to be "concise in conception", the editing time is dynamically adjusted. The adjustment formula is as follows: , where is the amount of time reduction in each video segment. is the reference threshold of video complexity index, is the video complexity index, is the maximum amount of editing time reduction allowed; The adjusted clip time Applied to the preset editing cycle to improve editing efficiency while ensuring video quality. The formula is as follows: , where is the adjusted editing period, It is the preset editing cycle; In the process of dynamically shortening the editing time, AI continuously monitors the progress of video editing and adjusts the results through real-time evaluation to ensure that the video quality is not affected. If there is any risk of quality degradation during the editing process, the feedback mechanism is triggered and corrections are made through the following formula: , where is the real-time feedback adjustment value, is the real-time calculated value of the current video complexity, It is to correct the editing progress factor to ensure that the quality remains stable during the editing process.

[0017] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention intelligently evaluates the complexity of the video and dynamically adjusts the editing time, especially for "simple-concept" videos, and speeds up the entire editing process by shortening the editing cycle. This optimization step effectively avoids the problem of waste of computing resources and low production efficiency caused by excessive editing time in videos with simpler content. By dynamically shortening the editing time, not only the video production process is accelerated, but also the overall production efficiency is greatly improved, ensuring that the market marketing activities can be responded to in a timely manner under the high concurrency demand of the e-commerce platform, avoiding missing key promotion opportunities. This is particularly important for e-commerce platforms, especially during peak promotion periods (such as Double Eleven, Black Friday, etc.), which can ensure that advertising materials are quickly launched to maximize sales and market competitiveness.

[0018] For videos with "complex ideas", the present invention continues to use the preset editing cycle to perform fine optimization and adjustment to ensure the maximum video quality. This process can accurately adjust the video content through real-time monitoring and intelligent evaluation mechanisms, combined with the long short-term memory network (LSTM) to deeply analyze features such as emotional fluctuations and visual contrast, to ensure that the video achieves the desired effect in terms of logic, fluency and visual impact. Through this intelligent editing method, the rhythm and picture content of the video can be accurately controlled to maintain high quality standards in video creation, thereby enhancing the user's viewing experience. Especially in the field of e-commerce, the quality of advertising videos directly affects the conversion rate. High-quality videos can better attract target audiences and enhance brand image and product appeal. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0020] Figure 1 The present invention is a method flow chart of the AI ​​material generation acceleration method for e-commerce promotion. DETAILED DESCRIPTION

[0021] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0022] The present invention provides Figure 1 The AI ​​material generation acceleration method for e-commerce promotion shown includes the following steps: At the beginning of video editing, AI technology first processes the video to be edited through a pre-set editing cycle to ensure that the initial stage of video editing meets the quality standards; The specific steps of AI technology processing the video to be edited through a pre-set editing cycle are as follows: In the initial stage of video editing, AI technology is used to analyze the overall content of the video to be edited; This includes basic parameters such as the video's duration, number of frames, resolution, and audio features. This analysis helps the system fully understand the basic structure of the video and provides a preliminary framework and direction for subsequent editing. Through this step, AI can quickly scan the visual and auditory information of the video to ensure that no important content is missed at the starting point of the editing.

[0023] Based on the preset editing cycle, AI technology will apply standardized editing rules to process video content; These rules include the minimum and maximum duration of shots, transition effects of scene changes, image and audio synchronization, etc., to ensure that the video follows consistent quality standards when editing. This stage ensures the basic consistency of video editing and lays a solid foundation for subsequent optimization.

[0024] Based on the application of editing rules, AI technology begins to automatically edit videos; This process will perform simple cutting, splicing and scene transitions according to the preset parameters of the video content, remove unnecessary parts, and adjust the transition effects. Although the goal of this stage of editing is to ensure maximum quality, the process still maintains a certain degree of flexibility to adapt to the needs of different video content.

[0025] The pre-set editing cycle refers to a fixed time range or processing time set in advance at the beginning of video editing, based on industry standards, the characteristics of the video content and quality requirements, for preliminary editing and processing of the video. During this cycle, AI will conduct a comprehensive analysis and optimization of the video to ensure that the logic, fluency, visual impact and other factors of the content are fully considered during the editing process, so that the video quality meets the expected standards. Usually, the pre-set editing cycle is longer to ensure that every part of the video can be processed in detail to avoid missing some important details or affecting the overall effect of the video due to time constraints. In a longer cycle, AI can more accurately adjust each shot, transition effect, audio matching, etc. of the video to ensure the coordination and refinement of all elements, thereby maximizing the quality of the video. Especially in complex or demanding video content, a longer editing cycle can ensure a balance between quality and effect.

[0026] The preset editing cycle covers the basic parameters of editing, such as the minimum and maximum duration of each shot, the type and duration of transition effects, etc. The main purpose of this stage is to ensure the overall logic and fluency of the video before conducting complex analysis, laying the foundation for subsequent optimization steps.

[0027] During the video editing process, AI will monitor and obtain various parameter information of the video content in real time, dynamically capture the real-time data of the video content, and provide a basis for complexity assessment; These parameters include, but are not limited to, picture brightness, color saturation, scene change frequency, motion detection, sound intensity, and the presence of subtitles and images. By acquiring these parameters in real time, the system can fully understand the dynamic changes and content characteristics of the video, providing detailed data support for subsequent complexity analysis. This step ensures that AI can flexibly respond to instant changes in video content during the editing process, ensuring the accuracy and adaptability of editing.

[0028] Preprocess the acquired data information, purify and normalize the data to facilitate subsequent feature extraction and analysis; The acquired raw video parameter information usually contains noise and irrelevant data, so preprocessing is required. The preprocessing steps include data cleaning, normalization, and dimensionality reduction. Specifically, the system filters out irrelevant or redundant information, adjusts the scale of the data to eliminate the dimensional differences between different parameters, and reduces the complexity of the data through dimensionality reduction techniques. This process ensures that the subsequent feature extraction can efficiently and accurately identify the key features that reflect the complexity of the video, thereby improving the effectiveness of the overall analysis.

[0029] Extract key indicators reflecting video complexity from preprocessed data, analyze the extracted indicators in detail within the detection window, and comprehensively and dynamically quantify the video complexity through the analyzed features; Key indicators reflecting the complexity of the video are extracted from the preprocessed data. The extracted features include the fluctuation degree of emotional expression in the video and the contrast between the picture elements in the video. Under the detection window, the extracted fluctuation degree of emotional expression in the video and the contrast between the picture elements in the video are analyzed to generate emotional fluctuation reference values ​​and visual contrast reference values ​​respectively. The complexity of the video is quantified by the emotional fluctuation reference values ​​and the visual contrast reference values.

[0030] The low degree of fluctuation in emotional expression in the video usually indicates that the video content is concisely conceived, because frequent changes in emotional fluctuations are often the manifestation of complex conception. Videos with concise conception tend to convey clear and direct information, and avoid excessive fluctuations or conflicts in emotional expression. Simple and direct emotional transmission can make it easier for the audience to understand the core content and purpose of the video without being disturbed by excessive emotional changes. In such videos, emotional expression is usually smooth and coherent, without overly intense ups and downs or multi-level emotional changes. This concise emotional fluctuation conforms to the simplified narrative structure, which can help the audience focus more on the core theme of the video and avoid unnecessary emotional burden or complexity. Excessive emotional fluctuations and transitions usually require more complex arrangements and deep emotional levels, and belong to video content with more complex conception. It can be seen that the degree of emotional fluctuation directly affects the overall complexity of the video conception. Low emotional fluctuations usually mean that the video content remains simple and clear in emotional expression, which conforms to the characteristics of concise conception.

[0031] The specific steps of analyzing the fluctuation degree of emotional expression in the video under the detection window to generate the reference value of emotional fluctuation are as follows: First, the speech and visual information in the video is processed through the emotion recognition algorithm to extract the emotional features of each frame in the video. For each frame, the facial expressions, voice intonation and visual content in the video are analyzed by multimodal emotion through AI technology to construct the emotion feature vector. The construction formula is as follows: , where is a visual emotion feature, indicating Emotional features extracted from video content (such as facial expressions, body language, movements, etc.) in the frame, is the audio emotion feature, indicating The emotional features obtained through speech analysis in the frame, is the text sentiment feature, indicating The emotional features extracted from the frame by text analysis, , as well as are weight coefficients, Visual emotional features In the sentiment feature vector The weight coefficient in determines the influence of visual signals (such as facial expressions, movements, body language, etc.) in the video on the overall emotional expression. is the audio emotion feature In the sentiment feature vector The weight coefficient in controls the contribution of the speech signal (such as tone, speaking speed, volume, etc.) in the video to the emotional expression. is the text sentiment feature In the sentiment feature vector The weight coefficient in determines the contribution of the text content in the video (such as subtitles, narration, dialogue, etc.) to the emotional expression. is the sentiment feature vector, indicating The comprehensive reflection of all emotional features in the frame video; The goal of this step is to obtain the emotion vector of each frame, which provides a basis for subsequent emotion fluctuation analysis.

[0032] After extracting the emotional features of each frame, the changing trend of the emotion is analyzed, and the local difference method is used to quantify the amplitude of the emotional change. and , calculate the change range of emotional features, the calculation expression is as follows: , where is the emotional feature vector of the next frame, is the fluctuation range control factor, It is the first Frame and the magnitude of emotional change between frames; Fluctuation range control factor It is an exponential coefficient used in the calculation of the amplitude of emotional fluctuations, which controls the sensitivity and amplification effect of the amplitude of emotional changes. Specifically, The role of is to weight and amplify this difference. Usually, The value of is set to a number greater than 1 (such as 2, 3, etc.) in order to enhance the significance of larger emotional fluctuations and ensure that larger emotional changes have a more prominent impact on the overall fluctuation assessment results. In this way, larger emotional fluctuations will be amplified, reflecting stronger emotional changes, while smaller fluctuations will be suppressed. The system can flexibly control the way in which emotional fluctuations are expressed in sentiment analysis and more accurately capture the emotional fluctuation patterns in the video, especially in video contents of different complexities, ensuring that the expression of emotional fluctuations matches the actual conception complexity of the video.

[0033] This step quantifies the emotional fluctuations between adjacent frames, emphasizing the changes of large fluctuations in exponential form, thereby avoiding information loss caused by smoothing. It aims to accurately capture the details of emotional changes in the video and provide necessary quantitative data for the subsequent calculation of fluctuation reference values.

[0034] After calculating the emotional fluctuation amplitude of each pair of adjacent frames, the frequency and intensity of the overall emotional fluctuation are evaluated. In order to better understand the overall trend of emotional fluctuation, the fluctuation amplitude of each pair of frames is accumulated through the weighted cumulative function, while considering its frequency of occurrence. The evaluation formula is as follows: , where is the reference value of emotional fluctuation, indicating that Next, the comprehensive calculation from the frame To the subsequent frame The degree of emotional fluctuations between It is the first Frame and The magnitude of emotional change between frames, is the time decay factor, which is an exponential function used to adjust the current frame With reference frame The time interval between The intensity of fluctuations To weight, is the attenuation coefficient, which is a positive constant that controls the rate of the time attenuation factor and adjusts the time weight of the emotional fluctuation between frames. is the total number of frames in the video; The formula used is Rather than , because it is necessary to comprehensively consider the reference frame The emotional fluctuations between the beginning and all subsequent frames. Specifically: Only the reference frame is calculated Its adjacent frames The emotional changes between the two reflect a single, local fluctuation. Although it can describe the reference frame The video can only capture the direct emotional changes of the video, but it cannot reflect the emotional fluctuation trend and complexity of the entire video.

[0035] In a cumulative summation process, the reference frame is calculated in sequence To end frame All adjacent frames of the emotion change. The formula can capture the emotional fluctuation amplitude of each pair of adjacent frames in the video and assign the distance reference frame to it under the action of the time decay factor. Emotional fluctuations at different time points have different weights. This accumulation method can fully quantify the To begin with, there is the emotionally volatile nature of the entire video.

[0036] use Rather than This is to make the formula not only focus on the reference frame It can not only take into account the local emotional changes of a video, but also comprehensively consider the emotional fluctuation intensity of all subsequent frames, reflecting the global trend of emotional changes and the time attenuation effect. This processing method is more in line with the complexity requirements of evaluating video emotional fluctuations.

[0037] This weighted accumulation method can balance the frequency and amplitude of emotional fluctuations, avoiding misleading information caused by a single emotional change that is too frequent or too strong, thereby accurately measuring the overall emotional fluctuations of the video.

[0038] The extracted emotional fluctuation information is integrated through a nonlinear function to generate an emotional fluctuation reference value. The purpose of this process is to determine the complexity of the overall emotion of the video based on the intensity and frequency of emotional fluctuations and quantify it into a numerical indicator. The generation formula is as follows: , where It is the reference value of emotional fluctuation. is a nonlinear adjustment parameter.

[0039] This step uses a nonlinear function to convert the accumulated fluctuation reference value Convert to a normalized emotional fluctuation reference value. Reference value The smaller the reference value, the smaller the emotional fluctuation of the video and the simpler the conception; conversely, the larger the reference value, the greater the emotional fluctuation of the video and the complex the conception.

[0040] Nonlinear tuning parameters It is the key parameter in the formula to control the degree of nonlinear compression. Its function is to adjust the cumulative emotional fluctuation intensity to the final reference value. Specifically, The value of determines the exponential function in the denominator The compressive strength of When the value of is large, the cumulative fluctuation intensity The influence of was significantly weakened, resulting in the reference value of emotional fluctuations The growth of tends to be gentle, which helps limit the extreme impact of highly volatile videos on the results; When the value is small, the index compression effect is weakened, and the intensity of emotional fluctuations more directly affects the emotional fluctuation reference value. , making the emotional fluctuation reference value more sensitive to changes in fluctuations. The main function of is to make the formula appropriately sensitive and balanced to video contents with different fluctuation amplitudes through nonlinear adjustment, thus ensuring the generation of reasonable and distinguishable reference values ​​in videos with simple and complex concepts.

[0041] The smaller the emotional fluctuation reference value generated by analyzing the fluctuation degree of emotional expression in the video under the detection window, the more concise the conception of the current video content is. Because videos with concise conception usually use a stable and unified emotional expression method, the emotional fluctuation is small, and the information conveyed by the video is more direct and clear, avoiding too many emotional transitions and complex emotional levels. On the contrary, videos with larger emotional fluctuation reference values ​​usually show stronger emotional fluctuations or multiple emotional transitions. Such videos often require more complex conception and design to show rich emotional levels and changes, which also indicates that the video content is more complex.

[0042] Low contrast between elements in a video usually indicates that the video content is simple in conception. Low contrast means that the differences between colors, brightness, and image elements in the video are smaller, and the picture presents a more unified and soft visual effect. This visual style is often consistent with video content with simple conception, because simple conception often pursues clear and direct information transmission, and avoids overly complex and gorgeous visual design. Low-contrast images usually reduce the visual burden of the audience, making the video content clearer and easier to understand, and focusing on the information transmission itself rather than complex visual effects. For example, in advertising or educational videos, simple content pays more attention to straightforward and clear expression, and usually chooses a relatively simple and soft color tone to ensure that the audience can focus on the core information of the video instead of being distracted by strong visual impact. In addition, low contrast can also make the video picture more harmonious and stable, avoid too much visual conflict, and thus enhance the overall simplicity and fluency of the video. In general, low-contrast images reduce unnecessary visual elements, making the overall style of the video simple and refined, which conforms to the design principle of simple conception.

[0043] The specific steps for analyzing the contrast between the elements in the video under the detection window to generate the visual contrast reference value are as follows: First, the video frames are preprocessed to extract key information from the video, including denoising and edge enhancement for each frame to improve the clarity and distinction of the image elements. After preprocessing, the brightness and chromaticity information of each pixel is extracted, and the contrast features of each pixel area are extracted using local contrast. The local contrast of each pixel is obtained by performing convolution operations on the local area. The formula is as follows: , where Indicates that in the video frame, the position The local contrast of the pixel at is the pixel coordinate, It's a pixel The brightness value represents the brightness intensity of the pixel. and is the offset of the convolution window, which is used to specify the position of adjacent pixels, and The value range is arrive , which means that around the current pixel is the neighborhood area of ​​radius, is the radius of the convolution window, which controls the range of the neighborhood when calculating local contrast, that is, for a given pixel , through the surrounding Analyze the pixels to calculate the local contrast; Convolution Window is a mathematical tool used for local feature extraction in image processing. In an image, a convolution window is usually a small matrix, usually with the center pixel as a reference, and operates on the entire image by sliding the window. Each time the window covers a small area in the image, the weighted sum of the pixel values ​​in the area and the corresponding weights in the window is calculated. The size and shape of the convolution window will affect the delicacy of the extracted features. Its main function is to extract specific features in the image, such as edges, textures, contrast, etc., through local operations. In the local contrast analysis of videos or images, the convolution window is used to calculate the brightness difference of the area around each pixel, and then quantify the contrast of the area. Through this process, the convolution window can effectively capture image details and local changes, help identify the visual differences between different areas in the image, and thus provide a basis for subsequent image analysis, classification, compression or enhancement operations.

[0044] The purpose of this step is to extract the local contrast features of each area in the video, quantify the visual differences between each area in the video, and highlight its detail levels.

[0045] Next, by calculating the contrast difference between adjacent areas in the video frame, the visual impact of the overall video is quantified. In order to accurately capture the contrast changes between the elements of the picture, the contrast difference factor is introduced. The calculation formula of the contrast difference factor is as follows: , where is the total number of pixels, It's a pixel The local contrast value, is the contrast difference factor; This step quantifies the contrast difference between the elements in the picture by calculating the difference between the contrast of the current pixel and its neighbors. The larger the value, the greater the difference between the picture elements and the more complex the visual effect. This step is used to capture the contrast differences between different areas and further analyze the visual impact of the video content.

[0046] Comprehensively analyze the spatial distribution in the video frame, especially the position and relationship of the elements in the picture, and use the spatial contrast factor to comprehensively evaluate the spatial contrast of the entire frame. The spatial contrast factor combines the spatial distribution information of the image and weights the contrast of each area. The evaluation formula is as follows: , where is the spatial contrast factor, is the spatial weight factor, which indicates the relative importance of the pixel in the overall spatial contrast calculation; By using the spatial contrast factor , which can better evaluate the spatial contrast distribution and thus reveal whether the video has high visual complexity.

[0047] Comprehensive pixel local contrast , contrast difference factor and the spatial contrast factor Generate visual contrast reference value, the generation formula is as follows: , where is the visual contrast reference value, is the contrast difference factor The weight coefficient of is the inter-contrast factor The weight coefficient of It is the global brightness contrast adjustment coefficient, which is used to balance the impact of local contrast on the overall visual contrast.

[0048] This step combines the comprehensive evaluation of contrast difference and spatial contrast with the global information of brightness and chromaticity to ensure that the reference value can fully reflect the visual contrast of the video. The smaller the value, the simpler the conception of the video content; the larger the value, the more complex the conception of the video.

[0049] The smaller the visual contrast reference value, which is generated by analyzing the contrast between the screen elements in the video under the detection window, the simpler the video content is. A low visual contrast value means that the differences between the screen elements in the video are small, the color and brightness present a unified and harmonious state, and there is no strong visual impact or abruptness. This usually reflects the simplicity of the video content design, which aims to convey clear and direct information, avoid excessive visual complexity, and ensure that the audience can easily follow the main idea of ​​the video content. Conversely, when the visual contrast reference value is large, the screen elements in the video show strong contrast, which may include sharp color changes or light and dark differences, and the visual effect is more stimulating and complex. This high-contrast screen design often appears in videos with more complex concepts, which may involve multi-level information transmission, rapid visual changes, or impactful scene design.

[0050] The quantified complexity index is input into the pre-trained long short-term memory network, and the video complexity is intelligently evaluated through the long short-term memory network to ensure the accuracy and reliability of the evaluation results; The emotional fluctuation reference value and visual contrast reference value generated after analysis are input into a pre-trained long short-term memory network, and a video complexity index is generated through the long short-term memory network. The video complexity is intelligently evaluated through the video complexity index.

[0051] The pre-trained long short-term memory network (LSTM) is a special type of recurrent neural network (RNN) that performs well when processing data with a temporal nature. Compared with traditional neural networks, LSTM can effectively solve the problem of long-term dependencies, that is, when processing sequence data, it can remember information over a longer time span without the gradient vanishing or gradient exploding like ordinary RNNs. Therefore, LSTM is widely used in natural language processing, speech recognition, and time series prediction. In the application of video editing, LSTM can capture the long-term dependencies of video content by analyzing the temporal changes of each frame in the video, especially when processing time series data such as emotional fluctuations and visual contrast, it can provide more accurate prediction and analysis. Through this temporal analysis, LSTM can identify complex temporal features such as the emotional change pattern of the video and the contrast change of visual elements, and then evaluate the overall complexity of the video.

[0052] In this scenario, the pre-trained LSTM network refers to a network that has been trained with a large amount of video data and has learned how to intelligently generate a video complexity index based on features such as the emotional fluctuations and visual contrast of the video. These training data usually include video clips of different complexities, and the network learns the relationship between various video features and video complexity through these data. When the emotional fluctuation reference value and the visual contrast reference value are input into the LSTM network, the network, through its internal loop structure, will infer based on the patterns in the historical data to determine how the temporal features of the emotional changes and visual contrast of the current video work together to affect the complexity of the video. Through this intelligent evaluation, the LSTM can output a video complexity index based on the specific features of the video content. This index will be used to quantify the complexity of the video and help the system determine the video editing cycle and whether further optimization and adjustment are needed. Because LSTM has a strong temporal modeling capability, it can efficiently analyze the dynamic features of the video and provide accurate complexity evaluation, thereby achieving efficient video editing and content optimization.

[0053] The long short-term memory network is not limited here, and can realize the reference value of emotional fluctuations and visual contrast reference values Perform comprehensive analysis to generate video complexity index In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; Video Complexity Index The generation formula is as follows: , where Emotional fluctuation reference values and visual contrast reference values The preset scaling factor of Both are greater than 0.

[0054] It can be seen from the video complexity index that the smaller the emotional fluctuation reference value generated after analyzing the fluctuation degree of emotional expression in the video under the detection window, the smaller the visual contrast reference value generated after analyzing the contrast between the picture elements in the video under the detection window, which means that the video complexity index generated when the video complexity is intelligently evaluated by the pre-trained long short-term memory network is smaller, indicating that the current video complexity is lower, and vice versa, it indicates that the current video complexity is higher.

[0055] Based on the evaluation results of the long short-term memory network, the video content is divided into two categories: "simple conception" and "complex conception"; The video complexity index generated by the pre-trained long short-term memory network when intelligently evaluating the video complexity is compared with the pre-set video complexity index reference threshold to divide the video content. The division steps are as follows: If the video complexity index is greater than or equal to a preset video complexity index reference threshold, the current video content is classified as complex in conception; If the video complexity index is less than a preset video complexity index reference threshold, the current video content is classified as having a simple concept.

[0056] Simple video content usually refers to videos that are relatively direct and concise in creative expression and visual presentation. Such videos usually adopt a relatively smooth and direct narrative structure, without too many twists or complex plot arrangements in the content. Complex video content is richer and more varied in creativity and visual presentation. Such videos usually involve complex narrative structures, with more plot levels, frequent camera switching, and more diverse combinations of visual elements. Emotional expression may experience more dramatic fluctuations, and the contrast between the picture and the audio is also more distinct. Complex videos bring a stronger impact and sense of participation to the audience through constantly changing visuals, sounds, and emotional fluctuations.

[0057] For videos classified as "complex in conception", the preset editing cycle will continue to be used to ensure maximum video quality and to conduct more detailed optimization and adjustments; For videos classified as "complex in conception", the preset editing cycle continues to be used to ensure maximum video quality, and more detailed optimization and adjustment are carried out in order to fully handle the complex elements in the video and ensure that every detail achieves the best effect. Complex videos usually involve more scene changes, dynamic images, and diverse visual or sound elements. These contents require more sophisticated editing to maintain the logic and coherence of the video, and avoid visual confusion or unclear information transmission due to too fast editing or insufficient adjustment. By extending the editing time and optimizing the editing process, AI can carefully adjust the transition, rhythm and emotional expression of each shot to ensure that the overall effect of the video meets the intended creative goals and maximizes its appeal and influence, ultimately improving the expressiveness of the video and the audience's acceptance.

[0058] For videos classified as "succinctly conceived", the editing time is dynamically shortened based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining editing quality; For videos classified as "simple in conception", the specific steps for dynamically shortening the editing time based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining the editing quality are as follows: After the video is judged to be "concise in conception", the editing time is dynamically adjusted. This adjustment is based on the difference between the complexity index of the video and the reference threshold. The adjustment formula is as follows: , where is the amount of time reduction in each video segment. is the reference threshold of video complexity index, is the video complexity index, is the maximum amount of editing time reduction allowed; The purpose of this step is to Shorten editing time. The smaller it is, the greater the reduction in editing time.

[0059] The adjusted clip time Applied to the preset editing cycle to improve editing efficiency while ensuring video quality. The formula is as follows: , where is the adjusted editing period, It is the preset editing cycle; This adjustment ensures that the editing process can be completed at a more efficient speed while maintaining video quality.

[0060] In the process of dynamically shortening the editing time, AI continuously monitors the progress of video editing and adjusts the results through real-time evaluation to ensure that the video quality is not affected. If there is any risk of quality degradation during the editing process, the feedback mechanism is triggered and corrections are made through the following formula: , where is the real-time feedback adjustment value, is the real-time calculated value of the current video complexity, It is to correct the editing progress factor to ensure that the quality remains stable during the editing process.

[0061] This feedback mechanism ensures that, despite the shortened editing time, the video quality remains within the predetermined high standards.

[0062] For videos classified as "simple in conception", the step of dynamically shortening the editing time based on the preset editing cycle is intended to improve the overall efficiency of video production while ensuring that the video quality is not affected. The core purpose of this process is to maximize production efficiency by reducing unnecessary editing time while ensuring that the final video effect meets the expected standards. Since "simple in conception" videos usually have lower complexity, manifested as fewer scene changes, lower action density and simplified visual elements, the traditional preset editing cycle is often lengthy and inefficient for such videos. By dynamically shortening the editing time, the system can complete the editing task more efficiently and avoid wasting too much time on those parts that are not complex and do not require fine-tuning, thereby saving computing resources and speeding up production progress.

[0063] In addition, dynamic adjustment of editing time can be intelligently optimized according to the actual complexity of the video content. For relatively concise videos, AI can automatically identify parts of the video that do not require too much detailed adjustment, thereby shortening the processing time of these parts. By precisely adjusting the time required for each editing stage, not only can the overall quality of the video be ensured to be unaffected, but the video production process can also be made more flexible, adapting to the high-concurrency, multi-tasking e-commerce environment, and responding to market demand in a timely manner. Ultimately, this dynamic editing time adjustment strategy not only improves efficiency, but also ensures the integrity and attractiveness of the video content, helping the platform to gain an advantage in the fierce market competition and quickly capture the attention of users.

[0064] The present invention intelligently evaluates the complexity of the video and dynamically adjusts the editing time, especially for "simple-concept" videos, and speeds up the entire editing process by shortening the editing cycle. This optimization step effectively avoids the problem of waste of computing resources and low production efficiency caused by excessive editing time in videos with simpler content. By dynamically shortening the editing time, not only the video production process is accelerated, but also the overall production efficiency is greatly improved, ensuring that the market marketing activities can be responded to in a timely manner under the high concurrency demand of the e-commerce platform, avoiding missing key promotion opportunities. This is particularly important for e-commerce platforms, especially during peak promotion periods (such as Double Eleven, Black Friday, etc.), which can ensure that advertising materials are quickly launched to maximize sales and market competitiveness.

[0065] For videos with "complex ideas", the present invention continues to use the preset editing cycle to perform fine optimization and adjustment to ensure the maximum video quality. This process can accurately adjust the video content through real-time monitoring and intelligent evaluation mechanisms, combined with the long short-term memory network (LSTM) to deeply analyze features such as emotional fluctuations and visual contrast, to ensure that the video achieves the desired effect in terms of logic, fluency and visual impact. Through this intelligent editing method, the rhythm and picture content of the video can be accurately controlled to maintain high quality standards in video creation, thereby enhancing the user's viewing experience. Especially in the field of e-commerce, the quality of advertising videos directly affects the conversion rate. High-quality videos can better attract target audiences and enhance brand image and product appeal.

[0066] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0067] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0068] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0069] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0070] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0071] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0072] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0073] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0074] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0075] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An AI material generation acceleration method for e-commerce promotion, characterized in that: The following steps are involved: At the beginning of video editing, AI technology first processes the video to be edited through a pre-set editing cycle to ensure that the initial stage of video editing meets the quality standards; During the video editing process, AI will monitor and obtain various parameter information of the video content in real time, dynamically capture the real-time data of the video content, and provide a basis for complexity assessment; Preprocess the acquired data information, purify and normalize the data to facilitate subsequent feature extraction and analysis; Extract key indicators reflecting video complexity from preprocessed data, analyze the extracted indicators in detail within the detection window, and comprehensively and dynamically quantify the video complexity through the analyzed features; The quantified complexity index is input into the pre-trained long short-term memory network, and the video complexity is intelligently evaluated through the long short-term memory network to ensure the accuracy and reliability of the evaluation results; Based on the evaluation results of the long short-term memory network, the video content is divided into two categories: "simple conception" and "complex conception"; For videos classified as "complex conception", the preset editing cycle will continue to be used to ensure maximum video quality and perform more detailed optimization and adjustments; For videos classified as "simple in conception", the editing time is dynamically shortened based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining editing quality.

2. The AI ​​material generation acceleration method for e-commerce promotion according to claim 1 is characterized in that: The specific steps of AI technology processing the video to be edited through a pre-set editing cycle are as follows: In the initial stage of video editing, AI technology is used to analyze the overall content of the video to be edited; Based on the preset editing cycle, AI technology will apply standardized editing rules to process video content; Based on the application of editing rules, AI technology begins to automatically edit videos.

3. The AI ​​material generation acceleration method for e-commerce promotion according to claim 1 is characterized in that: Key indicators reflecting the complexity of the video are extracted from the preprocessed data. The extracted features include the fluctuation degree of emotional expression in the video and the contrast between the picture elements in the video. Under the detection window, the extracted fluctuation degree of emotional expression in the video and the contrast between the picture elements in the video are analyzed to generate emotional fluctuation reference values ​​and visual contrast reference values ​​respectively. The complexity of the video is quantified by the emotional fluctuation reference values ​​and the visual contrast reference values.

4. The AI ​​material generation acceleration method for e-commerce promotion according to claim 3 is characterized in that: The specific steps of analyzing the fluctuation degree of emotional expression in the video under the detection window to generate the reference value of emotional fluctuation are as follows: First, the speech and visual information in the video is processed through the emotion recognition algorithm to extract the emotional features of each frame in the video. For each frame, the facial expressions, voice intonation and visual content in the video are analyzed by multimodal emotion through AI technology to construct the emotion feature vector. The construction formula is as follows: , where is a visual emotion feature, indicating Emotional features extracted from video content in frames, is the audio emotion feature, indicating The emotional features obtained through speech analysis in the frame, is the text sentiment feature, indicating The emotional features extracted from the frame by text analysis, , as well as are weight coefficients, is the sentiment feature vector, indicating The comprehensive reflection of all emotional features in the frame video; After extracting the emotional features of each frame, the changing trend of the emotion is analyzed, and the local difference method is used to quantify the amplitude of the emotional change. and , calculate the change range of emotional features, the calculation expression is as follows: , where is the emotional feature vector of the next frame, is the fluctuation range control factor, It is the first Frame and The magnitude of emotional change between frames.

5. The AI ​​material generation acceleration method for e-commerce promotion according to claim 4 is characterized in that: After calculating the amplitude of the emotional fluctuation of each pair of adjacent frames, the frequency and intensity of the overall emotional fluctuation are evaluated. The evaluation formula is as follows: , where It is the reference value of emotional fluctuation. It is the first Frame and The magnitude of emotional change between frames, is the time decay factor, is the attenuation coefficient, is the total number of frames in the video; The extracted emotional fluctuation information is integrated through nonlinear functions to generate emotional fluctuation reference values. The generation formula is as follows: , where It is the reference value of emotional fluctuation. is a nonlinear adjustment parameter.

6. The AI ​​material generation acceleration method for e-commerce promotion according to claim 3 is characterized in that: The specific steps for analyzing the contrast between the elements in the video under the detection window to generate the visual contrast reference value are as follows: First, the video frame is preprocessed to extract the key information in the video. After preprocessing, the brightness and chromaticity information of each pixel is extracted, and the contrast features of each pixel area are extracted using local contrast. The formula is as follows: , where Indicates that in the video frame, the position The local contrast of the pixel at is the pixel coordinate, It's a pixel The brightness value, and is the offset of the convolution window, which is used to specify the position of adjacent pixels, is the radius of the convolution window; Next, by calculating the contrast difference between adjacent areas in the video frame, the visual impact of the overall video is quantified. In order to accurately capture the contrast changes between the elements of the picture, the contrast difference factor is introduced. The calculation formula of the contrast difference factor is as follows: , where is the total number of pixels, It's a pixel The local contrast value, is the contrast difference factor.

7. The AI ​​material generation acceleration method for e-commerce promotion according to claim 6 is characterized in that: Comprehensively analyze the spatial distribution in the video frame, and use the spatial contrast factor to comprehensively evaluate the spatial contrast of the entire frame. The evaluation formula is as follows: , where is the spatial contrast factor, is the spatial weight factor; Comprehensive pixel local contrast , contrast difference factor and the spatial contrast factor Generate visual contrast reference value, the generation formula is as follows: , where is the visual contrast reference value, is the contrast difference factor The weight coefficient of is the inter-contrast factor The weight coefficient of is the global brightness contrast adjustment factor.

8. The AI ​​material generation acceleration method for e-commerce promotion according to claim 3 is characterized in that: The emotional fluctuation reference value and visual contrast reference value generated after analysis are input into a pre-trained long short-term memory network, and a video complexity index is generated through the long short-term memory network. The video complexity is intelligently evaluated through the video complexity index.

9. The AI ​​material generation acceleration method for e-commerce promotion according to claim 8 is characterized in that: The video complexity index generated by the pre-trained long short-term memory network when intelligently evaluating the video complexity is compared with the pre-set video complexity index reference threshold to divide the video content. The division steps are as follows: If the video complexity index is greater than or equal to a preset video complexity index reference threshold, the current video content is classified as complex in conception; If the video complexity index is less than a preset video complexity index reference threshold, the current video content is classified as having a simple concept.

10. The AI ​​material generation acceleration method for e-commerce promotion according to claim 9 is characterized in that: For videos classified as "simple in conception", the specific steps for dynamically shortening the editing time based on the preset editing cycle to speed up the editing process and improve efficiency while maintaining the editing quality are as follows: After the video is judged to be "concise in conception", the editing time is dynamically adjusted. The adjustment formula is as follows: , where is the amount of time reduction in each video segment. is the reference threshold of video complexity index, is the video complexity index, is the maximum amount of editing time reduction allowed; The adjusted clip time Applied to the preset editing cycle to improve editing efficiency while ensuring video quality. The formula is as follows: , where is the adjusted editing period, It is the preset editing cycle; In the process of dynamically shortening the editing time, AI continuously monitors the progress of video editing and adjusts the results through real-time evaluation to ensure that the video quality is not affected. If there is any risk of quality degradation during the editing process, the feedback mechanism is triggered and corrections are made through the following formula: , where is the real-time feedback adjustment value, is the real-time calculated value of the current video complexity, It is to correct the editing progress factor to ensure that the quality remains stable during the editing process.

Images