Method and system for generating clothing digital mannequin and optimizing parameters

Abstract

The invention provides a garment digital mannequin generation and parameter optimization method and system, and is applied to the technical field of data processing. Multi-platform environment parameters and clothing digital human core requirements are received, a cross-platform compatible mode is entered, an adaptation rule and basic configuration are generated, core assets are pre-stored, clothing and human body data are collected in the compatible mode, and effective features and parameters are extracted through screening. Then, according to a model generation rule, through multi-platform adaptation verification, an initial model and adaptation parameters are generated through parameter mapping, and an optimization mode is switched; an adaptive scene is defined in combination with multi-platform performance and a display standard, targeted optimization operation is formulated and finally converted into key information such as an actual adjustment action, a model precision optimization parameter and a cross-platform scheduling rule, and the multi-platform adaptation effect is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for generating and optimizing parameters of a digital clothing mannequin. Background Technology

[0002] With the rapid development of the digital economy and e-commerce industry, digital clothing mannequins have been widely used in various scenarios such as e-commerce displays, virtual try-ons, and live-streaming sales due to their efficient and flexible display advantages. However, significant differences exist between different application platforms in terms of hardware performance, software environment, and display standards, leading to core technical challenges for multi-platform adaptation of digital clothing mannequins.

[0003] In existing technologies, the generation and optimization of digital clothing mannequins are mostly developed for single platforms, lacking a unified cross-platform adaptation system, which mainly has the following problems: Insufficient multi-platform compatibility: The performance indicators and display requirements of different platforms vary greatly. The existing model's accuracy parameters, surface area, resource consumption and other configurations are difficult to meet the running requirements of multiple platforms at the same time. This often results in poor display effects on high-configuration platforms and lag or functional failures on low-computing-power platforms.

[0004] Lack of standardization in data collection and processing: The collection dimensions of clothing data (pattern, fabric, texture, etc.) and basic human body parameters are not consistent, and the data screening and verification process lacks cross-platform adaptation guidance, resulting in low utilization of effective data and difficulty in supporting the accurate generation of multi-platform models.

[0005] The model optimization lacks specificity: the existing optimization schemes do not fully take into account the characteristics and needs of different platforms, the parameter adjustment logic is vague, and no core module protection mechanism has been established, which may result in damage to core functions or failure to meet the adaptation requirements after optimization.

[0006] The lack of cross-platform scheduling and effect verification mechanisms results in poor stability of the model when migrating between multiple platforms, and difficulty in quantifying and continuously iterating the adaptation effect.

[0007] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0009] According to one aspect of this application, a method for generating and optimizing parameters of a digital human model for clothing is provided, comprising: receiving multi-platform system environment parameters and core generation requirements of the digital human model for clothing; entering a cross-platform compatibility mode according to preset adaptation standards; generating basic adaptation rules and mode settings; and pre-storing digital human basic model assets and core configuration parameters for multi-platform interface communication; in the cross-platform compatibility mode, collecting multi-dimensional clothing data and basic human body parameters; classifying and managing the data according to preset multi-source data filtering rules and extracting effective information; and generating clothing feature data and basic human body parameter information that conform to the adaptation standards; receiving model generation rule instructions; extracting core modeling elements; and, after multi-platform adaptation verification, optimizing parameters... The number mapping algorithm generates an initial clothing digital human model and basic adaptation parameters, then exits compatibility mode and switches to model optimization mode. Based on multi-platform performance requirements and display standards, the initial model information and basic adaptation parameters are processed to generate target platform adaptation scene information. The target platform adaptation scene information, initial model information, and mode switching result information are processed to generate targeted optimization operation information for the clothing digital human model. According to the preset optimization goals and multi-platform compatibility requirements, the optimization operation information is transformed into actual adjustment actions, and classification-related information is extracted to generate model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information, and model-platform adaptation mapping information.

[0010] Another aspect of this application is a system for generating and optimizing parameters of a digital human model for clothing, used to execute the executable instructions to perform the above-described method for generating and optimizing parameters of a digital human model for clothing.

[0011] According to another aspect of this application, an electronic device includes: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described method for generating and optimizing the parameters of a digital human model in clothing by executing the executable instructions.

[0012] According to another aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a second processor, implements the above-described method for generating and optimizing parameters of a digital human model for clothing.

[0013] This application provides a method and system for generating and optimizing digital clothing models. Focusing on multi-platform digital clothing model generation and parameter optimization, the core of this application achieves cross-platform adaptation through a three-stage process of "compatibility-generation-optimization." First, it receives multi-platform environment parameters and generation requirements, enters a cross-platform compatibility mode, collects multi-dimensional data on clothing and the human body, and generates a standardized dataset after screening and verification. Next, it extracts modeling elements, verifies multi-platform compatibility, generates an initial model and basic adaptation parameters through a parameter mapping algorithm, and switches to optimization mode. Finally, considering platform performance and display standards, it identifies core modules, matches optimization schemes, transforms these into adjustment actions, and outputs key information such as precision optimization parameters and cross-platform scheduling rules, forming a closed-loop process.

[0014] This application solves the multi-platform adaptation and compatibility challenges by unifying adaptation rules and dynamically adjusting parameters, avoiding issues such as poor display on high-configuration platforms and lag on low-computing-power platforms, thus ensuring stable operation across multiple platforms. Standardized data collection and processing workflows improve effective data utilization, supporting accurate model generation, while a core module protection mechanism prevents optimization from damaging core functions. Clearly defined cross-platform scheduling rules and effect verification standards enable quantitative evaluation and continuous iteration of adaptation effects, significantly improving the adaptation efficiency and display quality of digital clothing mannequins in e-commerce displays, virtual try-on scenarios, and other applications.

[0015] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0016] Figure 1 This document illustrates a flowchart of a method for generating and optimizing the parameters of a digital human model for clothing, provided in an embodiment of this application. Figure 2 The diagram shows a structural schematic of a system for generating and optimizing digital clothing models according to an embodiment of this application. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] In one implementation, Figure 1 A schematic flowchart illustrating a method for generating and optimizing parameters of a digital human model for clothing according to an embodiment of this application is shown.

[0019] S101 receives multi-platform system environment parameters and core generation requirements of clothing digital human models, enters cross-platform compatibility mode according to preset adaptation standards, generates adaptation basic rules and mode settings, and pre-stores digital human basic model assets and multi-platform interface communication core configuration parameters.

[0020] In one implementation, two types of core input information are received to lay the data foundation for cross-platform adaptation and model generation: multi-platform system environment parameters cover the hardware performance indicators of each target platform, such as running memory, computing power limit, graphics processing capability, software environment configuration, such as operating system type and version, supported rendering engines, interface compatibility standards, network transmission characteristics, such as data transmission bandwidth, latency threshold and other key parameters, to fully adapt to the operating conditions of different platforms.

[0021] The core requirements for generating digital human models for clothing are used to define the application scenarios of digital human models, such as e-commerce display, virtual try-on, and live streaming sales; style positioning, such as realistic style, cartoon style, and minimalist style; functional requirements, such as flexibility of body movements, accuracy of clothing material reproduction, and real-time interactive response speed; and quantitative indicators, such as the upper limit of model face count, rendering frame rate requirements, and cross-platform synchronization latency threshold.

[0022] The system activates a cross-platform compatibility mode based on a preset adaptation standard. This standard is formulated based on industry-standard cross-platform technical specifications and the adaptation requirements of various mainstream platforms, covering data format compatibility standards, model performance adaptation thresholds, interface communication protocol specifications, and more. The system automatically detects input multi-platform environment parameters and matches the corresponding compatibility mode configuration to ensure that subsequent data acquisition, model generation, and parameter optimization processes meet the basic requirements for multi-platform collaborative operation, eliminating technical barriers between different platforms.

[0023] Based on the received input information and the core requirements of cross-platform compatibility, two key categories of content are generated: Clearly defined data collection format standards, such as using JSON format for clothing data and standardized numerical formats for human body parameters; data accuracy requirements, such as retaining clothing size parameters with an accuracy of 0.1cm and human skeletal angles with an accuracy of 1°; priority rules for model adaptation, such as prioritizing smooth model operation on low-computing-power platforms and prioritizing improved visual effects on high-configuration platforms; and security specifications for interface data interaction, such as data encryption transmission protocols and permission verification rules.

[0024] Define the data processing flow under compatibility mode, such as the order of data acquisition, filtering, verification, and storage; the stage division of model generation, such as the stage division of basic model construction, parameter adaptation, and optimization adjustment; and the triggering mechanism for cross-platform communication, such as the automatic synchronization triggering when data is updated and the adaptation verification triggering after model adjustment.

[0025] Simultaneously, the pre-deposit of two types of core resources was completed to provide direct support for subsequent processes: Standardized digital human basic model resources were organized and stored, including a basic human skeleton model, which covers skeletal structure templates for different genders and age groups, clearly defining key parameters such as the number of skeletal nodes and the range of joint motion; a general clothing pattern basic model, such as the pattern framework for basic categories like tops, pants, and skirts, defining key size parameters and adjustment ranges; and a basic material library, containing basic physical property parameters of common clothing fabrics, such as gloss, roughness, and elasticity coefficient. All assets are stored in a cross-platform compatible format, supporting quick access and secondary adjustments.

[0026] Organize the key parameters for interface communication between various target platforms, including data transmission protocols such as HTTP, HTTPS, and TCP / IP; interface call specifications such as API version, request method, and call frequency limits; data interaction formats such as JSON and XML; and permission verification standards such as token verification, key verification, and data transmission timeout thresholds. Build a standardized interface configuration library to ensure efficient and stable communication between the system and various platforms.

[0027] S102, in cross-platform compatibility mode, collects multi-dimensional clothing data and basic human body parameters, classifies and manages them according to preset multi-source data filtering rules, extracts effective information, and generates clothing feature data and basic human body parameter information that meet the adaptation standards.

[0028] In one implementation, under cross-platform compatibility mode, multi-dimensional clothing data covering pattern, fabric, texture, and accessory attributes, along with basic parameters of human skeleton and body shape, are collected. Under cross-platform compatibility mode, two types of core data are acquired through multi-source data acquisition methods to ensure data coverage of all modeling needs: for the pattern, key dimensional parameters such as garment length, shoulder width, waist circumference, and skirt hem curvature are collected; for the fabric, physical property parameters such as density, thickness, elasticity, and breathability are collected; for the texture, visual feature parameters such as pattern style, color distribution, and texture resolution are collected; and for accessories, related parameters such as accessory type, size specifications, material properties, and connection position with the garment are collected. Collection methods include 3D scanning, material testing instrument measurement, and image recognition extraction to ensure that the data accurately reflects the actual characteristics of the clothing.

[0029] Focusing on the human skeleton and body shape, skeletal structural parameters such as skeletal node coordinates, joint range of motion, and bone length are collected; body shape dimensional parameters such as height, weight, shoulder width, chest circumference, waist circumference, hip circumference, and limb circumference are also collected. These parameters are collected using professional tools such as 3D human scanning equipment and motion capture equipment to ensure that the accuracy meets the adaptation requirements of digital human modeling.

[0030] Based on pre-defined multi-source data filtering rules, the collected clothing data and basic human body parameters are categorized, archived, and redundant information is removed. A hierarchical classification algorithm is used, dividing the data into six major categories according to data type: clothing pattern, fabric attribute, texture feature, accessory association, human skeleton, and human body shape. Each category is further subdivided into secondary categories based on specific parameter dimensions. For example, the clothing pattern category is subdivided into secondary categories such as top pattern, trouser pattern, and skirt pattern; the human body shape category is subdivided into secondary categories such as torso dimension and limb dimension, thus establishing a structured data archiving system.

[0031] A data redundancy detection algorithm is used to identify and remove duplicate data (such as repeated size parameters collected multiple times for the same garment), invalid data (such as outliers outside the reasonable range, incomplete data lacking key information), and redundant related data (such as accessory and decorative details data unrelated to garment modeling). By setting key parameters such as data similarity threshold and reasonable range threshold, valid data is automatically filtered and retained, reducing the burden of subsequent data processing.

[0032] Extract core features of clothing and key dimensions of the human body. Use a feature importance assessment algorithm to select features that play a decisive role in the display effect and suitability of the clothing. For example, core features of the garment's pattern include the size proportions of key parts and the curves of the garment's outline; core features of the fabric include attributes that affect the wearing effect, such as elasticity, luster, and drape; core features of the texture include key visual parameters such as the main color tone of the pattern, the repeating period of the texture, and clarity; and core features of accessories include related features such as the size and material compatibility with the clothing.

[0033] Based on the human adaptation requirements of digital human modeling, we extract dimensional data that plays a key role in clothing fit and body movement display. For example, key skeletal dimensions include the length of the core bones supporting the clothing and the limit angle of joint movement; key body shape dimensions include core circumferences such as chest, waist, and hip circumference that determine the fit of the clothing, as well as basic dimensions such as height and shoulder width, to ensure that the extracted parameters can accurately support the adaptation modeling of clothing and the human body.

[0034] The extracted features and parameters are validated for compliance according to cross-platform adaptation standards to ensure that the data format and accuracy meet modeling requirements. Data formats are verified for uniformity and standardization based on cross-platform data interaction standards. For example, whether numerical parameters use standardized numerical formats, whether text parameters conform to preset description specifications, and whether image texture data is in a cross-platform compatible image format. Data that does not meet the standards is automatically converted using format conversion algorithms to ensure smooth data transmission and retrieval across different platforms.

[0035] Set accuracy thresholds for each parameter and use accuracy verification algorithms to check whether the data accuracy meets the standards. For example, the accuracy of clothing size parameters must meet ±0.1cm, the accuracy of human body circumference parameters must meet ±0.5cm, and the texture resolution must meet the preset pixel threshold. For data that does not meet the accuracy standards, process it through data completion, re-collection, etc., to ensure that the data accuracy can support the accurate display of digital human models on multiple platforms.

[0036] Considering the technical characteristics of each target platform, verify whether the data meets the platform's operational requirements. For example, low-computing-power platforms have constraints on the number of model faces and the amount of data, so it's necessary to verify whether the extracted feature data can be processed efficiently on that platform; different rendering engines have different requirements for material parameters, so it's necessary to verify whether the fabric property parameters are compatible with the rendering engines of each platform. Through compatibility testing algorithms, proactively identify and address any adaptation risks of data in cross-platform applications.

[0037] A standardized dataset of clothing features and a set of basic human body parameters are generated. Using a standardized data encoding format, the core features of clothing are stored in a structured manner according to a classification system. Each data entry includes metadata information such as feature name, feature value, data precision, collection method, and verification results, forming a uniformly sized dataset of clothing features to ensure data traceability and usability.

[0038] Similarly, standardized coding is used to integrate key human body dimension data, which is stored in categories according to skeletal structure and body shape. Each parameter is associated with corresponding accuracy indicators, verification conclusions and other auxiliary information to generate a structured human body basic parameter information set, providing standardized input for the construction of digital human body models.

[0039] Output a list of valid data after classification, the results of adaptation standard verification, and the final core information on clothing features and human body parameters. Output three types of key information to provide clear data support for subsequent model generation: According to the aforementioned six major data categories and secondary classifications, sort out and output a detailed list of valid data, clarifying key information such as parameter names, data values, data status (qualified), and storage paths under each data category, so as to facilitate quick query and retrieval during subsequent modeling.

[0040] Outputting a structured report, it includes format verification results, accuracy verification results, and compatibility verification results for each parameter, marking data that failed verification, the reasons, and the handling methods, clearly presenting the overall data verification process. It integrates the core content of the clothing core feature dataset and the human body basic parameter information set, extracting key data summaries to form the final core information of clothing features and human body parameters, which directly serves as the core data input for digital human model modeling.

[0041] S103 receives the model generation rule instruction, extracts the core modeling elements, and after multi-platform adaptation verification, generates the initial clothing digital human model and basic adaptation parameters through the parameter mapping algorithm, and exits the compatibility mode to switch to the model optimization mode.

[0042] In one implementation, an initial strategy is generated by receiving business parameters, budget rules, and historical data from the footwear and apparel industry. This strategy constructs a basic algorithmic architecture for traffic filtering, dynamic bidding, and scenario-channel adaptation through multi-sub-model collaboration. Budget control benchmarks are set using weighted allocation and time series analysis algorithms. A brand category attribute database and a cross-channel interface configuration library are pre-stored. The system receives three types of core input information, builds a multi-sub-model collaborative architecture, and pre-stores core resources to provide foundational support for subsequent decisions: receiving business parameters from the footwear and apparel industry (covering category type, product positioning, target customer attributes, seasonal adaptation characteristics, etc.), budget rules (including total budget amount, time-period consumption limit, unit cost constraint, channel budget allocation ratio limit, etc.), and an initial strategy (including traffic filtering direction, bidding benchmark range, channel placement priority suggestions, etc.) generated based on historical campaign data (traffic data, bidding records, conversion effects, budget execution status, etc.).

[0043] This system establishes a collaborative infrastructure encompassing a traffic filtering and adaptation analysis model, a dynamic bidding algorithm model, and a scenario-channel adaptation algorithm model. The traffic filtering and adaptation analysis model comprises a data access layer, a tag matching layer, and a value assessment layer, responsible for connecting traffic with category attributes, calculating audience tag fit, and initially determining traffic value. The dynamic bidding algorithm model consists of a feature extraction layer, a value verification layer, and a bid calculation layer, extracting user behavior features, verifying traffic value, and calculating target bids. The scenario-channel adaptation algorithm model covers a scenario construction layer, a channel adaptation layer, and a creative matching layer, generating placement scenarios, analyzing channel fit, and defining creative matching rules. All models interact bidirectionally through standardized data interfaces, forming a closed-loop decision-making process.

[0044] A benchmark is established by combining a weighted allocation algorithm and a time series analysis algorithm. The weighted allocation algorithm uses the market demand for each category (quantifiable indicators such as search volume and sales growth rate), brand placement priority, and historical conversion performance (core indicators such as ROI and conversion rate) as dimensions, assigning different weights to calculate the budget allocation coefficient for each category and channel. The time series analysis algorithm combines the seasonal patterns and promotional periods of the footwear and apparel industry, analyzes historical data from the same period, and formulates daily and weekly budget consumption plans. The percentile method is used to set the maximum bid threshold for a single click and impression, clarifying the cost control boundaries.

[0045] A pre-stored brand category attribute database (integrating information such as category type, product positioning, target customer profile, and seasonal adaptability, and storing it in a standardized format) and a cross-channel interface configuration library (organizing data transmission protocols, interface call specifications, data interaction formats, and permission verification standards for each channel, and unifying data interaction standards).

[0046] This system integrates with omnichannel traffic bidding requests, focusing on target traffic through multi-condition intersection filtering. It utilizes a traffic filtering and adaptation analysis model to calculate matching weights, constructs a multi-dimensional evaluation matrix to filter high-value traffic, and then uses a traffic value completion model to address gaps, generating bidding opportunities and traffic value information tailored to the budget. It integrates with omnichannel traffic bidding requests (including traffic source, exposure time, user profile, traffic type, etc.) and pre-stored brand category attributes, introducing footwear and apparel audience tagging rules (age, spending power, style preference, purchase frequency, etc.) and a budget adaptation filtering mechanism (maximum cost per traffic, time-based budget, etc.). Employing a multi-condition intersection filtering algorithm, it first matches category suitability, then filters the target customer group, and finally eliminates traffic exceeding budget constraints, focusing on the target traffic range and generating preliminary traffic value assessment results (value scores based on category suitability, audience matching, and budget suitability).

[0047] Build a traffic filtering and adaptation analysis model, associate the focused traffic data with the attributes and characteristics of footwear and apparel categories and the matching standards of audience tags, calculate the traffic matching weight through a tag fit calculation model (using a weighted summation algorithm, assigning weights according to the degree of influence of tags on conversion), and establish a dynamic association mechanism of category-audience-budget (automatically increase the matching weight threshold of high-value audience tags when the budget is insufficient).

[0048] Using traffic matching weight as the core dimension, and integrating category attributes, audience tags, and budget constraints, a multi-dimensional traffic screening and evaluation matrix is ​​constructed (row dimensions represent individual traffic, and column dimensions represent matching weight, category fit, tag completeness, and budget suitability). The analytic hierarchy process is used to determine the weight percentage of each column dimension, and a comprehensive score threshold is set to screen out high-value traffic.

[0049] The traffic value completion model addresses gaps in traffic quantity or quality. This model comprises a gap identification layer (analyzing gap types), a feature completion layer (completing missing features based on similar historical traffic data), and a value assessment layer (performing a secondary value assessment of the completed traffic). Combined with a budget-adaptive optimization mechanism, when the budget is sufficient, the assessment thresholds for non-core dimensions are appropriately reduced, while the weight of key traffic features is strengthened. The model integrates precise screening results with qualified traffic after completion to generate a budget-adapted bidding opportunity list (including traffic ID, placement channel, and suggested bidding range) and traffic value information to be assessed (including comprehensive value score, conversion potential prediction, and budget adaptation instructions).

[0050] User behavior features are extracted, and traffic value is verified using PCTR / PCVR models. A multi-dimensional bidding decision feature set is generated, and the target bid is calculated using a budget smoothing algorithm. Combined with real-time bidding scenario adaptability analysis, the system switches to real-time delivery mode and outputs bidding results and mode confirmation information. Pre-set dynamic bidding algorithm instructions (including bidding strategy type, adjustment trigger conditions, and value verification standards) are mapped and matched with the traffic value information to be evaluated to extract user behavior features (browsing duration, click frequency, historical purchase records, add-to-cart behavior, etc.). Traffic value is verified using PCTR / PCVR prediction models. The models learn the mapping relationship between features and conversion probabilities based on historical conversion data, outputting click and conversion prediction values ​​and a value verification score.

[0051] Based on the conversion patterns in the footwear and apparel industry, the analytic hierarchy process (AHP) was used to determine the weight percentage of each feature (historical purchase records 0.3, add-to-cart behavior 0.25, click frequency 0.2, browsing duration 0.15, traffic value verification result 0.3, of which pCTR prediction value accounts for 0.4 and PCVR prediction value accounts for 0.6). The data of each feature was normalized to the 0-1 range by a normalized weighted summation algorithm, and the weighted score was calculated according to the weight to form a multi-dimensional bidding decision feature set.

[0052] Based on a multi-dimensional bidding decision feature set, a budget smoothing algorithm is used to calculate the target bid. The formula is: Target Bid = Base Bid × (Sum of Weighted Scores of Each Feature × Bid Adjustment Coefficient + Budget Fit Coefficient). The base bid is set based on the average conversion cost of similar historical traffic. The bid adjustment coefficient is dynamically set according to the feature weighted scores (1.2 for scores above 0.8, 1.0 for scores between 0.6 and 0.8, and 0.8 for scores below 0.6). The budget fit coefficient is set according to the remaining percentage of the current budget (1.1 for scores above 80%, 1.0 for scores between 50% and 80%, and 0.9 for scores below 50%). The algorithm verifies whether the target bid is within the preset bid limit, predicts the traffic acquisition volume and budget consumption rate, and ensures that it conforms to the budget consumption rhythm.

[0053] The system quantifies and scores bids based on three dimensions: real-time bidding intensity, bidding levels within the same product category, and channel traffic price fluctuations. A weighted sum is then used to obtain a comprehensive suitability score (≥0.7 indicates good suitability). This score is then compared to the predicted ROI value corresponding to the bid to determine if it meets a preset threshold. Once confirmed, the system switches to real-time execution mode, outputting dynamic bidding results (target bid, adjustment basis, feasibility verification conclusion) and mode switch confirmation information (switch time, execution status, feedback mechanism).

[0054] By integrating bidding results and traffic value assessment conclusions, a unified advertising environment is built. A scenario-channel adaptation algorithm determines the granularity of channel adaptation and the rules for matching creative materials, marking the priority and bidding logic of each channel, and forming an operational plan for creative material distribution and channel matching. Advertising delivery instructions (including delivery goals, delivery period, target audience, etc.) and cross-channel interface configuration parameters are imported into the advertising scenario building system. Dynamic bidding results and traffic value assessment conclusions are integrated, and a data fusion algorithm eliminates data format differences and redundancy. An environment building algorithm constructs an integrated workflow for scenario adaptation, bidding matching, and execution scheduling, achieving data interoperability and logical linkage among all stages.

[0055] Based on the differences in footwear and apparel category attributes and channel traffic characteristics, a scenario-channel adaptation algorithm is selected. First, the matching degree between category attributes (style, price range, seasonality) and channel traffic characteristics (user age distribution, spending power, browsing habits) is quantified. Then, the channel adaptation granularity is determined based on the matching degree (full-scale delivery to core channels, precise delivery to potential channels, and selective delivery to alternative channels). Finally, material matching rules are generated in combination with the needs of the delivery scenario (dynamic materials adapted to short video channels, text and image materials adapted to search engine channels, etc.).

[0056] For omnichannel advertising scenario information (including the type of each channel, target customer segmentation, traffic scale prediction, etc.), combined with dynamic bidding results and mode switching confirmation information, a multi-dimensional information association and labeling algorithm is used to quantitatively evaluate the priority of each channel (core, important, general) from three dimensions: expected campaign performance, budget adaptability, and traffic quality. The algorithm clarifies the material adaptation standards (format, content focus, size requirements) and bidding execution logic (adjusting trigger conditions, optimizing frequency, and upper limit control rules) for each channel.

[0057] By integrating scenario-channel matching rules, bidding execution requirements, and material adaptation standards across three dimensions—"channel, creative material, and bid"—this approach clarifies the distribution ratio, placement time, and bid execution details for different creative materials across various channels. It also incorporates budget control requirements and a real-time feedback mechanism to form a directly executable operational plan for advertising material distribution and channel matching. This plan includes a creative material distribution list, placement ratio, time slot arrangement, bid details, performance monitoring indicators, and optimization and adjustment processes.

[0058] The system links campaign objectives, budget constraints, and operational plans, identifies configurations with insufficient adaptability and clarifies optimization directions, quantifies and organizes campaign objective priority and budget allocation data, generates feasibility analysis reports and budget efficiency trend charts, and ultimately integrates and outputs decision-making information including content adaptation, budget scheduling, and performance attribution. It associates pre-set campaign objectives (brand promotion, new product traffic generation, sales improvement, etc., and corresponding quantitative indicators), budget control requirements, and advertising material distribution and channel matching operational plans, establishing a logical mapping relationship along the "campaign objective → budget constraint → channel matching rules → material adaptation standards" link to ensure a closed-loop execution logic. A multi-dimensional adaptability scoring algorithm is used to quantify scores from four dimensions: campaign objective fit, budget adaptability, traffic matching, and historical performance data (each dimension has a weight of 0.25, with a total score below 0.6 indicating insufficient adaptability). It identifies adaptability issues in channel combinations and material types (such as mismatch between channel traffic and target audience, or mismatch between material style and channel user preferences), and clarifies optimization directions (adjusting channel combination ratios, changing material types, etc.).

[0059] A standardized quantification algorithm is used to transform non-numerical information into quantifiable indicators, and numerical information is normalized. Prioritization of campaign targets is determined by weighted scoring based on strategic importance, expected ROI, and time urgency (weights 0.4, 0.3, and 0.3, respectively). Allocation coefficients are calculated based on historical performance, traffic volume, and target relevance for each channel to determine the budget allocation for each channel and time period. Operational parameters such as campaign time periods, bidding ranges, creative change frequency, and performance monitoring thresholds for each channel are quantified. Based on traffic attribute characteristics, the appropriate creative types, bidding strategies, and campaign frequencies are quantified, generating a list of data including campaign target priorities, budget allocation ratios, channel operation points, and traffic adaptation suggestions.

[0060] Integrating relevant information on ad placement configurations, core optimization directions, and quantitative list data, the system employs a SWOT-quantitative fusion algorithm. This algorithm combines strengths (highly adaptable channel combinations, high-quality creatives), weaknesses (insufficient adaptability of some channels), opportunities (target audience traffic growth), threats (intensified industry competition), and quantitative data (budget adequacy rate, probability of goal achievement) to calculate a feasibility score (≥0.7 indicates good). This generates an ad placement feasibility analysis report (including SWOT analysis, risk assessment, and countermeasures). Based on time series analysis algorithms, the system predicts budget consumption speed, ROI trends, and goal achievement rates across different time periods, generating a budget utilization efficiency trend chart to quantitatively demonstrate the evolution of the strategy's alignment with goals.

[0061] The system integrates core configuration optimization annotations, quantitative configuration lists, feasibility reports, and efficiency trend charts to generate multi-dimensional decision-making information. Content adaptation parameters are determined based on material matching rules and quantitative adaptation suggestions; budget scheduling rules are formulated in conjunction with budget allocation ratios, consumption trends, and flexible budget strategies; the performance attribution model employs a multi-touchpoint attribution algorithm (combining linear and location attribution), collecting data on placement and conversion across various channels to calculate the attribution value of each channel and material; channel-category association mapping information is generated based on channel adaptation granularity and category attribute relationships, ultimately outputting a complete decision result containing all the above information.

[0062] S104 processes the initial model information and basic adaptation parameters based on multi-platform performance requirements and display standards to generate target platform adaptation scenario information.

[0063] In one implementation, the accuracy parameters, face size, and basic adaptation parameters of the initial clothing digitizer model are extracted according to multi-platform performance indicators and display standards. A hierarchical parameter extraction algorithm is used to extract the core parameters of the initial clothing digitizer model based on multi-platform performance indicators and display standards, ensuring that the parameters cover key dimensions of adaptation: performance indicator-related parameters focus on the hardware performance and software limitations of multiple platforms, extracting the accuracy parameters, face size, and resource consumption parameters of the initial model. These parameters directly affect the smoothness of the model's operation on different platforms.

[0064] The display standard parameters are based on multi-platform display standards, extracting the model's visual presentation parameters and interaction adaptation parameters to ensure that the parameters meet the platform's display effect requirements.

[0065] The basic adaptation parameters are the basic parameters related to the model's adaptation to multiple platform interfaces and data interaction, including interface compatibility parameters and cross-platform synchronization parameters, which provide basic data for subsequent adaptation adjustments.

[0066] Based on platform characteristics and display requirements, the adaptation priority of the model on different platforms is clarified, and the direction of parameter adjustment and the core dimensions of scenario adaptation are determined. Based on platform characteristics and display requirements, the adaptation rules are clarified through a priority evaluation algorithm, and the core direction of parameter adjustment is determined: the hardware performance limits, software environment characteristics, and user scenarios of each target platform are analyzed to form a platform characteristic profile.

[0067] A multi-dimensional weighted evaluation algorithm is adopted, using platform user scale, business importance, and technical adaptation difficulty as evaluation dimensions, and assigning different weights to calculate the adaptation priority of each platform. At the same time, in accordance with the principle of "performance assurance first, display effect second", the priority of parameter adjustment is determined, with low computing power platforms prioritizing ensuring the smooth operation of the model, and high configuration platforms prioritizing improving the visual effect.

[0068] Based on platform characteristics and priorities, determine the direction of parameter adjustments. Clearly define the core dimensions of scenario adaptation, including performance adaptation, display adaptation, and interaction adaptation, ensuring that adjustments focus on core adaptation needs.

[0069] Define matching rules between model parameters and platform performance to ensure a balance between adaptability and display quality. Develop multi-dimensional matching rules to balance model adaptability and display quality, ensuring the rules are implementable and verifiable: performance adaptation rules define a quantitative matching relationship between model parameters and platform performance. For example, set an upper limit on the number of faces in the model based on the platform's computing power level; set a threshold for video memory usage based on the platform's memory capacity; and set an upper limit on the amount of data transferred from the model based on the platform's data transfer bandwidth, ensuring the model runs stably on the platform.

[0070] The display effect matching rules are based on the platform's display standards, setting the acceptable range for the model's visual parameters. For example, high-resolution platforms have higher texture precision, while ordinary resolution platforms can appropriately reduce texture precision; platforms supporting advanced rendering engines enable enhanced lighting and reflection parameters, while basic rendering engine platforms use simplified lighting and reflection parameters to ensure that the display effect meets the platform's capability limits.

[0071] The dynamic adjustment rules are based on the logic of dynamically adapting parameters and adjusting model parameters in conjunction with the platform's real-time performance status. For example, when the platform's CPU utilization exceeds a threshold, the model's polygon count is automatically reduced; when the platform is in a low-bandwidth network environment, the model's data transmission frequency is reduced to prioritize core visual effect parameters, balancing adaptation stability and display quality.

[0072] The classification and extraction results, adaptation priorities, and matching rules are integrated and processed to generate target platform adaptation scenario information, which includes platform performance adaptation parameters, display standard compliance requirements, and core scenario adaptation directions. A scenario-parameter-rule association algorithm is used to categorize the extracted model parameters by platform dimension and associate them with corresponding adaptation priorities and matching rules, forming a four-dimensional association system of "platform-parameter-priority-rule" to ensure a closed-loop information logic.

[0073] Three types of core information are generated: platform performance adaptation parameters, display standard compliance requirements, and core adaptation directions for the scenario. Following cross-platform data interaction standards, the integrated information is output in a structured format, including a list of adaptation parameters for each target platform, a priority comparison table, matching rule explanations, and adjustment direction guidance, providing clear and actionable scenario-based support for subsequent model optimization operations.

[0074] S105 processes the target platform adaptation scene information, initial model information, and mode switching result information to generate targeted optimization operation information for the clothing digital human model model.

[0075] In one implementation, the target platform adaptation scene information, initial clothing digitized human model information, and mode switching result information are standardized and integrated to generate a multi-dimensional optimized input dataset. A data standardization and integration algorithm is used to process and integrate the target platform adaptation scene information, initial clothing digitized human model information, and mode switching result information in a unified format to generate a multi-dimensional optimized input dataset. According to cross-platform data interaction standards, unstructured data of the three types of information is transformed into structured format, unifying data encoding, field naming, and parameter units to eliminate format differences. Performance adaptation parameters and display compliance requirements in target platform adaptation scenario information are organized according to the "platform-parameter-threshold" structure; precision parameters and face size of initial clothing digital human model information are standardized according to the "model dimension-parameter value-attribute" structure; and switching status and execution feedback of mode switching result information are unified according to the "mode-status-timestamp" structure.

[0076] By using a key field matching algorithm, a mapping relationship is established between three types of information, namely "platform identifier" and "model ID", to form a multi-dimensional optimization input dataset that includes platform adaptation requirements, current model status, and mode switching status. This ensures that the data logic is coherent and traceable, providing complete data support for subsequent optimization analysis.

[0077] The core differences in the dataset are extracted, including platform performance adaptation gaps, model accuracy shortcomings, and mode switching adaptation vulnerabilities, generating an optimization requirement feature vector. Using a difference analysis algorithm, core differences are extracted from the multi-dimensional input dataset and transformed into a quantified optimization requirement feature vector: focusing on three key difference dimensions: platform performance adaptation gaps refer to the difference between the current model parameters and the platform performance requirements (e.g., model polygon count exceeding the platform limit, accuracy parameters not meeting platform standards); model accuracy shortcomings refer to the model's own deficiencies in visual presentation, interactive response, etc. (e.g., insufficient texture clarity, delayed action response); and mode switching adaptation vulnerabilities refer to adaptation anomalies that occur during mode switching (e.g., model parameters incompatible with the platform after switching, functional failure). Through threshold comparison, trend analysis, and other methods, various differences are accurately identified and labeled.

[0078] The extracted core differences are quantified and encoded, with each difference corresponding to a feature dimension. The degree of difference is represented in numerical form (such as the proportion of faces exceeding the upper limit or the numerical difference of inaccuracy). A multi-dimensional optimization requirement feature vector is constructed to intuitively reflect the priority and core direction of model optimization.

[0079] The model's non-adjustable core modules, including the basic human skeletal structure and core clothing pattern logic, are locked, and a module protection mechanism is established. The basic human skeletal structure and core clothing pattern logic are identified as non-adjustable core modules. The basic human skeletal structure includes core parameters such as bone node distribution, joint range of motion, and bone proportions to ensure the rationality of the digital human's basic form. The core clothing pattern logic includes key clothing size proportions, pattern outline curves, and clothing-human fit rules to ensure basic clothing display adaptability. A parameter freezing algorithm is used to lock the relevant parameters of the core modules, preventing modification during the optimization process.

[0080] A module protection verification algorithm is constructed to verify in real time whether the operation involves core module parameters during the subsequent optimization scheme generation and execution process. If an illegal operation is detected, it will be automatically intercepted and an alarm will be triggered. At the same time, an operation log recording mechanism is established to record all optimization operations in a complete manner, which is convenient for traceability and rollback, and ensures the stability and security of the core module.

[0081] The optimization requirement feature vectors are correlated with the model's adjustable parameter library, and targeted optimization solutions are matched one by one according to platform adaptation priority. The adjustable parameter dimensions in the model are sorted out, including precision adjustment parameters (texture precision, vertex precision), resource optimization parameters (face size, memory usage), interaction adaptation parameters (response latency, motion smoothness), etc., and stored in a categorized manner according to parameter type, adjustment range, and related impact dimensions to form the model's adjustable parameter library.

[0082] Calculate the similarity between the feature vector of the optimization requirement and each parameter in the adjustable parameter library, and prioritize matching the parameters with the highest correlation to the core differences. According to the platform adaptation priority (core platform > important platform > alternative platform), match optimization solutions for the optimization requirements of each platform one by one, and clarify the adjustment parameters, adjustment direction (increase / decrease, expand / shrink) and adjustment range suggestions to ensure that the solution specifically addresses the differences.

[0083] The optimization scheme is validated for compliance based on a module protection mechanism to ensure that the core modules are not tampered with, and to generate targeted optimization operation information for the clothing digital human model. The protection mechanism's validation algorithm checks each optimization scheme to see if it involves modification of core module parameters. If the scheme does not touch the core modules and the adjusted parameters are within a safe range (e.g., face count adjustment does not fall below the basic threshold, precision adjustment does not affect the basic display effect), it is deemed compliant; if there is a risk of violation, it is returned for re-matching and adjustment. Simultaneously, the scheme is validated to ensure it meets the platform's adaptation requirements, ensuring that the adjusted parameters meet the performance and display standards of the target platform.

[0084] The compliant optimization solutions are transformed into standardized optimization operation information, which clarifies the execution object (parameter dimension), execution method (adjustment logic), execution threshold (adjustment range), execution order (sorted by priority), and expected effect (optimized achievement standard) of each operation, forming a list of directly executable optimization operations, providing clear guidance for the implementation of model optimization.

[0085] S106, based on the preset optimization goals and multi-platform compatibility requirements, transforms the optimization operation information into actual adjustment actions, extracts classification-related information, and generates model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information, and model-platform adaptation mapping information.

[0086] In one implementation, based on preset optimization goals and multi-platform compatibility requirements, core information such as model-specific optimization operation information, platform technical attributes, and initial model parameters is categorized and integrated to generate standardized model adjustment execution unit information. This categorization and integration follows a three-dimensional framework: "optimization operation - platform attributes - model parameters." Optimization operation information is categorized by adjustment type (accuracy optimization, resource optimization, interactive adaptation); platform technical attributes are categorized by hardware performance, software environment, and operating characteristics; and initial model parameters are categorized by accuracy parameters, polygon count, and adaptation parameters, ensuring clear information dimensions.

[0087] Unify the data format, encoding rules and field naming of various types of information, and associate and bind the execution logic of optimization operations, the constraints of platform attributes and the current state of initial model parameters to form a standardized model adjustment execution unit of "operation object - platform constraints - parameter benchmark - optimization target". Each unit corresponds to a specific optimization action independently, ensuring that the execution logic is clear and traceable.

[0088] Based on multi-platform adaptation standards and model accuracy requirements, the content layout of standardized model adjustment execution units is designed, clarifying the indicator proportions for model accuracy optimization, cross-platform adaptation, effect verification, and adaptation mapping, and generating content design information for adjustment units. Unit content is arranged according to the logic of "priority-core indicators-auxiliary indicators," with priorities divided according to platform importance and optimization urgency; core indicators focus on four dimensions: model accuracy optimization, cross-platform adaptation, effect verification, and adaptation mapping; auxiliary indicators include related information such as operation time and resource consumption changes, ensuring a clear distinction between primary and secondary content.

[0089] Through a multi-dimensional weighted evaluation algorithm, combined with multi-platform adaptation standards and model accuracy requirements, the following four core indicators are assigned weights: Model accuracy optimization accounts for 35%, focusing on key dimensions such as texture accuracy and vertex accuracy; Cross-platform adaptation accounts for 30%, covering parameter compatibility and interface adaptability; Effect verification accounts for 20%, including visual effect compliance and interaction response speed; Adaptation mapping accounts for 15%, linking the correspondence between model parameters and platform characteristics to ensure that optimization actions focus on core needs.

[0090] Based on the operating characteristics of different platforms and the requirements for achieving optimization results, model accuracy thresholds, cross-platform adaptation and compatibility rules, and optimization effect verification standards are set to ensure that adjustments meet the actual application requirements of multiple platforms. Accuracy quantification thresholds are set according to platform display standards and model application scenarios. These include texture accuracy thresholds, vertex accuracy thresholds, and motion smoothness thresholds, clearly defining the accuracy compliance range for different platforms and ensuring that the model's visual effects and interactive experience meet platform requirements.

[0091] Establish parameter compatibility rules to clarify the adaptation range of model parameters on different platforms (such as the number of faces and the upper limit of platform memory usage); interface adaptation rules to standardize the data interaction format and call frequency limits between the model and the platform interface; and runtime stability rules to set constraint thresholds such as CPU utilization and memory usage of the model on the platform to avoid adaptation conflicts.

[0092] Establish multi-dimensional verification standards. The visual effect standard clarifies the judgment criteria such as texture clarity and color reproduction. The performance effect standard sets the target values ​​for rendering frame rate and response latency. The compatibility effect standard specifies the functional consistency requirements of the model when running on multiple platforms, ensuring that the optimization effect is quantifiable and verifiable.

[0093] The process involves processing the integration results, content design information, and compliance rules of the model adjustment execution unit to generate model precision optimization parameter information, cross-platform adaptation scheduling rules, optimization effect verification information, and model-platform adaptation mapping information, including unit composition, verification indicators, and judgment specifications. This completes the optimization process for the digital human clothing model. A rule-parameter-effect association algorithm is employed to deeply integrate the integration results, content design information, and compliance rules of the model adjustment execution unit, constructing a three-dimensional linkage system of "action-indicator-constraint." Each optimization action in the execution unit is categorized by adjustment type and bound to one of the four core indicators in the content design, with weights allocated according to preset proportions. Precision adjustment actions are prioritized for association with model precision optimization indicators, while resource optimization actions are primarily linked to cross-platform adaptation indicators, ensuring a high degree of alignment between optimization actions and core objectives. A mapping relationship is established between the specific requirements of each core indicator and the constraints in the compliance rules. Model precision optimization indicators are associated with constraints such as texture precision thresholds and vertex precision thresholds, while cross-platform adaptation indicators are linked to parameter compatibility rules and interface adaptation rules, ensuring that the achievement of each indicator has a clear constraint basis. The algorithm iterates through all relationships to verify whether the optimization actions meet the target ratio requirements and whether the constraints can support the achievement of the target. If there are conflicts, the relationships are automatically adjusted or the conflict points are fed back to ensure that every decision has clear execution logic and feasible verification standards.

[0094] Based on the deep fusion results, four types of core decision information required for model optimization are generated, covering all dimensions of accuracy, adaptation, verification, and mapping: The accuracy standards for each target platform are clearly defined, including thresholds for key dimensions such as texture accuracy, vertex accuracy, and motion smoothness; the adjustment range for each accuracy parameter is determined, specifying the adjustment direction and specific numerical range; an optimized accuracy benchmark is set as a target reference for subsequent model adjustments, ensuring that the model accuracy on each platform meets the display and interaction requirements. Dynamic parameter adjustment logic is formulated, clarifying the real-time adjustment strategy for model parameters under different platform operating states; the adaptation process during platform switching is standardized, including parameter synchronization methods, interface compatibility verification steps, and state transition rules; resource allocation rules are established to rationally allocate computing power, memory, and other resources according to platform priority and optimization needs, ensuring the stability of multi-platform operation.

[0095] We have compiled a multi-dimensional set of verification metrics, covering visual effects, performance effects, and compatibility effects; clarified the verification methods for each metric, including the application of automated testing tools and key points for manual verification; and formulated compliance judgment standards, setting acceptable ranges and handling methods for each metric to ensure that optimization effects are quantifiable and verifiable. We have established a correspondence between model parameters and platform characteristics, clarifying the optimal configuration of model parameters for different platforms; prioritized adaptation based on platform importance and adaptation difficulty, guiding optimization resources towards core platforms; and established conflict resolution rules, clarifying priority judgment criteria and compromise solutions for potential parameter configuration conflicts during multi-platform adaptation to ensure optimal adaptation solutions.

[0096] All decision-making information is formatted according to cross-platform data interaction standards to form a unified and standardized optimization operation guide: Four types of decision-making information are organized using a structured format, with unified data encoding, field naming, and presentation formats to ensure smooth transmission and retrieval of information across different platforms and systems. Output includes detailed parameter configuration tables, rule description documents, and operation step guides, facilitating direct implementation by technical personnel. The optimization operation guide clearly defines the executing entity, execution order, and execution time limit for each decision-making information item, while also highlighting verification requirements and anomaly handling schemes for key nodes. Precision parameter adjustments must be performed in the order of "threshold verification - parameter modification - effect detection." If the detection fails to meet the standards, a secondary adjustment process is triggered. The output decision-making information serves as the direct basis for model adjustments, guiding implementation; simultaneously, data from the execution process is fed back to the initial information integration stage, providing data support for subsequent optimization iterations, forming a closed-loop process of "information integration - rule setting - decision output - implementation - data feedback - iterative optimization," ensuring continuous optimization of the model's multi-platform adaptability.

[0097] In one implementation, such as Figure 2 As shown, this application also provides a system for generating and optimizing parameters of a digital human model for clothing, including: The initialization configuration module 201 is used to receive multi-platform system environment parameters and core generation requirements of clothing digital human models, enter cross-platform compatibility mode according to preset adaptation standards, generate adaptation basic rules and mode settings, and synchronously pre-store digital human basic model assets and multi-platform interface communication core configuration parameters. The data acquisition and processing module 202 is used to collect multi-dimensional clothing data and basic human body parameters in cross-platform compatibility mode, classify and manage the data according to preset multi-source data filtering rules and remove redundant information, extract effective information and verify the compliance of cross-platform adaptation standards, and generate clothing feature data and basic human body parameter information that meet the requirements. The initial model generation module 203 is used to receive model generation rule instructions, extract core modeling elements, and after multi-platform adaptation verification, generate an initial clothing digital human model and basic adaptation parameters through a parameter mapping algorithm, exit the cross-platform compatibility mode and switch to the model optimization mode. The adaptation scenario construction module 204 is used to classify and extract initial model information and basic adaptation parameters and determine adaptation priority based on multi-platform performance requirements and display standards, set matching rules between parameters and platforms, and generate target platform adaptation scenario information. The optimization operation formulation module 205 is used to standardize and integrate the target platform adaptation scenario information, initial model information and mode switching result information, extract the core differences to generate optimization requirement feature vectors, combine the model core module protection mechanism, match targeted optimization solutions and complete compliance verification, and generate targeted optimization operation information for the clothing digital human model model. The model optimization implementation module 206 is used to transform optimization operation information into actual adjustment actions based on preset optimization goals and multi-platform compatibility requirements. It integrates model adjustment execution unit information, content design information and achievement rules, extracts classification-related information, and generates model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information, and model-platform adaptation mapping information to complete the multi-platform adaptation optimization of the digital human clothing model.

[0098] The computer-readable storage medium provided in the above embodiments of this application and the method for generating and optimizing digital human models of clothing provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.

[0099] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of the method, system, electronic device, and readable storage medium for evaluating the generation and parameter optimization of clothing digital human models are basically similar to the embodiments of the method for generating and optimizing clothing digital human models described above, and are therefore described relatively simply. Relevant parts can be referred to in the descriptions of the embodiments of the method for generating and optimizing clothing digital human models described above.

Claims

1. A method for generating and optimizing parameters of a digital human model for clothing, characterized in that, include: Receive multi-platform system environment parameters and core generation requirements of clothing digital human models, enter cross-platform compatibility mode according to preset adaptation standards, generate adaptation basic rules and mode settings, and pre-store digital human basic model assets and core configuration parameters for multi-platform interface communication. In cross-platform compatibility mode, multi-dimensional clothing data and basic human body parameters are collected, classified and managed according to preset multi-source data filtering rules, and effective information is extracted to generate clothing feature data and basic human body parameter information that meet the adaptation standards. Receive the model generation rule instructions, extract the core modeling elements, and after multi-platform adaptation verification, generate the initial clothing digital human model and basic adaptation parameters through the parameter mapping algorithm, and exit the compatibility mode to switch to the model optimization mode. Based on the performance requirements and display standards of multiple platforms, the initial model information and basic adaptation parameters are processed to generate target platform adaptation scene information; Process the target platform adaptation scenario information, initial model information, and mode switching result information to generate targeted optimization operation information for the clothing digital human model model; Based on the preset optimization goals and multi-platform compatibility requirements, the optimization operation information is transformed into actual adjustment actions, relevant classification information is extracted, and model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information, and model-platform adaptation mapping information are generated.

2. The method as described in claim 1, characterized in that, In cross-platform compatibility mode, multi-dimensional clothing data and basic human body parameters are collected, classified and managed according to preset multi-source data filtering rules, and effective information is extracted to generate clothing feature data and basic human body parameter information that meet the adaptation standards, including: In cross-platform compatibility mode, multi-dimensional clothing data covering pattern, fabric, texture, and accessory attributes, as well as basic parameters of human skeleton and body shape, are collected; Based on the pre-set multi-source data filtering rules, the collected clothing data and basic human body parameters are classified, archived, and redundant information is removed. Extract core features of clothing and key dimensions of the human body; The extracted features and parameters are verified for compliance according to cross-platform adaptation standards to ensure that the data format and accuracy meet the modeling requirements. Generate a standardized dataset of clothing features and a set of basic human body parameter information; Output a list of valid data after classification, the results of adaptation standard verification, and the final core information of clothing features and human body parameters.

3. The method as described in claim 1, characterized in that, Upon receiving model generation rule instructions, core modeling elements are extracted. After multi-platform adaptation verification, an initial clothing digital human model and basic adaptation parameters are generated through a parameter mapping algorithm. The system then exits compatibility mode and switches to model optimization mode, including: The system receives business parameters, budget rules, and historical data from the footwear and apparel industry to generate initial strategies. It then constructs a basic algorithm architecture for traffic filtering, dynamic bidding, and scenario-channel adaptation multi-sub-model collaboration. Combined with weighted allocation and time series analysis algorithms, it sets budget control benchmarks and pre-stores brand category attribute databases and cross-channel interface configuration libraries. Connect with omnichannel traffic bidding requests, focus on target traffic through multi-condition intersection filtering, use traffic filtering and adaptation analysis model to calculate matching weights, build a multi-dimensional evaluation matrix to filter high-value traffic, and then use traffic value completion model to process gaps to generate bidding opportunities and traffic value information that are suitable for the budget. Extract user behavior features, verify traffic value through PCTR / PCVR model, generate multi-dimensional bidding decision feature set, calculate target bid through budget smoothing algorithm, combine real-time bidding scenario adaptability analysis, switch to real-time delivery mode, and output bidding results and mode confirmation information. By integrating bidding results and traffic value assessment conclusions, an integrated advertising environment is built. Through scenario-channel adaptation algorithms, the granularity of channel adaptation and the rules for matching creative materials are determined. The priority and bidding logic of each channel are marked, and a plan for creative material distribution and channel matching is formed. The system links campaign objectives, budget constraints, and operational plans, identifies configurations with insufficient adaptability and clarifies optimization directions, quantifies and organizes campaign objective priority and budget allocation ratio data, generates feasibility analysis reports and budget efficiency trend charts, and finally integrates and outputs decision-making information including content adaptation, budget scheduling, and performance attribution.

4. The method as described in claim 1, characterized in that, Based on multi-platform performance requirements and display standards, the initial model information and basic adaptation parameters are processed to generate target platform adaptation scenario information, including: Based on multi-platform performance indicators and display standards, the accuracy parameters, face size, and basic adaptation parameters of the initial clothing digital human model are classified and extracted. Based on platform characteristics and display requirements, clarify the adaptation priority of the model on different platforms, and determine the direction of parameter adjustment and the core dimensions of scenario adaptation. Set matching rules between model parameters and platform performance to ensure a balance between adaptability and display effect; The classification and extraction results, adaptation priorities, and matching rules are integrated and processed to generate target platform adaptation scenario information that includes platform performance adaptation parameters, display standard compliance requirements, and core scenario adaptation directions.

5. The method as described in claim 4, characterized in that, The target platform adaptation scenario information, initial model information, and mode switching result information are processed to generate targeted optimization operation information for the clothing digital human model, including: Standardize and integrate the target platform adaptation scenario information, initial clothing digital human model information, and mode switching result information to generate a multi-dimensional optimized input dataset; Extract the core differences in the dataset, including platform performance adaptation gaps, model accuracy shortcomings, and mode switching adaptation vulnerabilities, and generate feature vectors for optimization requirements; Lock the model's non-adjustable core modules, including the basic human skeletal structure and core clothing pattern logic, and establish a module protection mechanism; The optimization requirement feature vector is associated with the model adjustable parameter library, and targeted optimization solutions are matched one by one according to the platform adaptation priority. The optimization scheme is validated for compliance based on the module protection mechanism to ensure that the core modules are not tampered with, and the digital mannequin model of clothing is generated for targeted optimization operation information.

6. The method as described in claim 4, characterized in that, Based on the preset optimization goals and multi-platform compatibility requirements, the optimization operation information is transformed into actual adjustment actions, classification-related information is extracted, and model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information, and model-platform adaptation mapping information are generated, including: Based on the preset optimization goals and multi-platform compatibility requirements, the core information of the model-specific optimization operation information, platform technical attributes, and initial model parameters is classified and integrated to generate standardized model adjustment execution unit information. Based on multi-platform adaptation standards and model accuracy requirements, the content layout of the standardized model adjustment execution unit is designed, clarifying the proportion of indicators for model accuracy optimization, cross-platform adaptation, effect verification, and adaptation mapping, and generating content design information for the adjustment unit. Based on the operating characteristics of different platforms and the requirements for achieving optimization results, we set model accuracy thresholds, cross-platform adaptation and compatibility rules, and optimization effect verification standards to ensure that the adjustment actions meet the actual application requirements of multiple platforms. The model adjustment execution unit integration results, content design information and compliance rules are processed to generate model accuracy optimization parameter information, cross-platform adaptation scheduling rule information, optimization effect verification information and model-platform adaptation mapping information, which include unit composition, verification indicators and judgment specifications, thus completing the optimization process of this clothing digital human model.

7. A system for generating and optimizing parameters of a digital human model for clothing, characterized in that, The system is used to execute executable instructions to perform the method for generating and optimizing the parameters of a digital human model for clothing as described in any one of claims 1 to 6.

8. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the method for generating and optimizing the parameters of a digital human model of clothing according to any one of claims 1 to 6 by executing the executable instructions.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the method for generating and optimizing the parameters of a digital human model of clothing as described in any one of claims 1 to 6.

Images