AI digital human construction method and system based on 3D modeling technology

Abstract

The invention provides an AI digital human construction method and system based on a 3D modeling technology, and belongs to the technical field of 3D modeling, and the method comprises the steps: obtaining digital human construction demand basis information of a vending machine; determining a digital human construction demand of the vending machine according to the digital human construction demand basis information; performing digital human construction design according to the digital human construction demand to obtain a design scheme; based on a 3D modeling technology, according to a design scheme, AI digital human construction is carried out. According to the AI digital human construction method based on the 3D modeling technology, the digital human construction demand basis information of the vending machine is obtained, the digital human construction demand is determined based on different digital human construction demand basis information, and the digital human construction design is performed based on the digital human construction demand to obtain the design scheme. The AI digital human construction is carried out based on the 3D modeling technology and the design scheme, the customized digital human can realize communication with the user, the method is more humanized, and the user experience of the vending machine is improved.

Description

Technical Field

[0001] The present invention relates to the field of 3D modeling technology, and in particular to an AI digital human construction method and system based on 3D modeling technology. Background Art

[0002] Digital Human is a digital image created through a variety of digital intelligence technologies such as modeling. It has human appearance, voice language, body movements or thinking functions. It is a digital intelligent entity (AI Agent) with learning, generation, interaction and other functions supported by a large model.

[0003] The application scope of digital people is very wide, including but not limited to entertainment, education, customer service, medical care, cultural tourism and other fields. It can be used as a virtual customer service or sales assistant to provide 24-hour uninterrupted service, lower operating costs, and no need for rest. The image, voice, and style of digital people can be customized according to the needs of the brand, making it more in line with the brand image and market positioning.

[0004] At the same time, with the development of artificial intelligence technology, vending machines can integrate more functions, such as facial recognition, voice interaction and data analysis, making vending machines more intelligent. Therefore, combining digital human technology with vending machines is a huge advantage for vending machines that need to operate for a long time. However, the vending devices that currently use virtual digital humans for vending generally use a fixed virtual digital human image for each user, and users cannot make personalized choices, resulting in a poor experience.

[0005] In view of this, there is an urgent need for an AI digital human construction method and system based on 3D modeling technology to at least solve the above-mentioned deficiencies. Summary of the invention

[0006] One of the purposes of the present invention is to provide an AI digital human construction method and system based on 3D modeling technology, obtain digital human construction requirement basis information of a vending machine, the digital human construction requirement basis information includes character image selection instructions actively input by the user and passively captured images, determine the digital human construction requirement based on different digital human construction requirement basis information, and perform digital human construction design based on the digital human construction requirement to obtain a design plan, construct an AI digital human based on the 3D modeling technology and the design plan, and the customized digital human can communicate with the user, which is more humane and improves the user experience of the vending machine.

[0007] The AI ​​digital human construction method based on 3D modeling technology provided in an embodiment of the present invention includes:

[0008] Obtaining the digital human construction requirement basis information of the vending machine, the digital human construction requirement basis information including: character image selection instructions and shot images;

[0009] Determine the digital human construction requirements for the vending machine based on the digital human construction requirements information;

[0010] Design the digital human construction according to the digital human construction requirements and obtain the design plan;

[0011] Based on 3D modeling technology, AI digital humans are constructed according to the design plan.

[0012] Preferably, obtaining the information based on the requirements for constructing a digital human of a vending machine includes:

[0013] Obtaining a character image selection instruction of a user who uses a vending machine to purchase goods;

[0014] If the acquisition is successful, the character image selection instruction is used as the basis information for the digital human construction requirements;

[0015] If the acquisition fails, an image of the user who uses the vending machine to purchase goods is obtained by a preset camera device of the vending machine, and the image is used as the basis information for the digital human construction requirements.

[0016] Preferably, according to the digital human construction requirement basis information, determining the digital human construction requirement of the vending machine includes:

[0017] The type of information on which the digital human construction needs to be based;

[0018] When the information type is a character image selection instruction, the character image selection instruction is parsed to obtain the digital human construction requirements;

[0019] When the information type is captured images, the digital human construction requirements are predicted based on the captured images.

[0020] Preferably, predicting the digital human construction requirements based on the captured images includes:

[0021] Obtain the user's historical construction records based on the captured images;

[0022] If the acquisition is successful, the digital human construction requirements are determined based on the historical construction records;

[0023] If the acquisition fails, based on the preset character portrait extraction template, the character portrait is extracted according to the captured image;

[0024] According to the historical figure image selection instructions, determine the historical figure portrait demand probability table;

[0025] According to the demand probability table of character portraits and historical character portraits, determine the digital human model with the highest demand probability corresponding to the historical character portrait that matches the person-household portrait;

[0026] Use the digital human model as the digital human construction requirement.

[0027] Preferably, the digital human construction design is performed according to the digital human construction requirements to obtain a design solution, including:

[0028] Based on the preset demand feature extraction template, the demand features of digital human construction are extracted according to the digital human construction requirements;

[0029] Obtain design solutions based on required features and the preset digital human material library.

[0030] The AI ​​digital human construction method based on 3D modeling technology provided in the embodiment of the present invention also includes:

[0031] Supplement the digital human material library;

[0032] Among them, the digital human material library is supplemented, including:

[0033] Based on the digital human resource big data platform, supplement the digital human material library;

[0034] and / or,

[0035] Upload materials based on local supplementary nodes to supplement the digital human material library.

[0036] Preferably, based on the digital human resource big data platform, the digital human material library is supplemented, including:

[0037] Based on the digital human resource big data platform, obtain the shared information of shared materials;

[0038] Extracting shared information features based on a preset shared information feature extraction template and shared information;

[0039] Determine the type of shared material heat quantification index according to the shared information characteristics; the shared material heat quantification index type includes: weight index and value index;

[0040] Determine the shared material heat quantification index value according to different shared material heat quantification index types;

[0041] Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index type is a weight index to obtain a target weight value;

[0042] Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index types are numerical indicators to obtain a target sum value;

[0043] Multiply the target weight value and the target sum value to obtain the thermal value of the shared material;

[0044] Compare the material similarity between the shared material with the largest heat value and the local real-time material of the digital human material library. If the material similarity is less than the preset material similarity threshold, the digital human material library is supplemented by connecting the sharing path according to the sharing information of the corresponding shared material.

[0045] Preferably, the digital human material library is supplemented based on the uploaded materials of the local supplementary node, including:

[0046] Get the user evaluation record of the vending machine;

[0047] Parse user evaluation records and obtain user evaluation semantics;

[0048] Determine the material evaluation record in the user evaluation record according to the user evaluation semantics and the preset material description semantics;

[0049] Expand the material evaluation records on a preset time axis based on the generation time, and obtain the time density of the material evaluation records within a preset time interval;

[0050] If the time density is greater than or equal to the preset time density threshold, the corresponding time interval is used as the target time interval;

[0051] Obtaining material evaluation features of each material evaluation record within a target time interval;

[0052] Perform feature clustering on material evaluation features;

[0053] If clustering is successful, obtain the feature clustering results;

[0054] Determine the need for additional materials based on feature clustering results;

[0055] The material supplementation requirement is sent to the local supplementation node, and the uploaded material corresponding to the material supplementation requirement uploaded by the local supplementation node is obtained to supplement the digital human material library.

[0056] Preferably, clustering is successful, including:

[0057] The target number of cluster members in the feature clustering is greater than the number of first evaluation records corresponding to the time density;

[0058] or,

[0059] The target number of cluster members in the feature cluster is greater than the number of cluster members in other feature clusters, and the target number of members is less than the first number of evaluation records corresponding to the time density and greater than the second number of evaluation records corresponding to the time density threshold.

[0060] The AI ​​digital human construction system based on 3D modeling technology provided by the embodiment of the present invention includes:

[0061] The information acquisition subsystem is used to acquire the digital human construction requirement basis information of the vending machine, and the digital human construction requirement basis information includes: character image selection instructions and shot images;

[0062] A digital human construction requirement determination subsystem is used to determine the digital human construction requirement of the vending machine according to the digital human construction requirement basis information;

[0063] The design scheme acquisition subsystem is used to design the digital human construction according to the digital human construction requirements and obtain the design scheme;

[0064] The digital human construction subsystem is used to construct AI digital humans based on 3D modeling technology and according to the design plan.

[0065] The beneficial effects of the present invention are:

[0066] The present invention obtains the digital human construction requirement basis information of the vending machine, the digital human construction requirement basis information includes the character image selection instruction actively input by the user and the photographed image passively taken, determines the digital human construction requirement based on different digital human construction requirement basis information, performs digital human construction design based on the digital human construction requirement to obtain the design scheme, constructs the AI ​​digital human based on the 3D modeling technology and the design scheme, and the customized digital human can communicate with the user, which is more humane and improves the user experience of the vending machine.

[0067] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the present application documents.

[0068] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0070] Figure 1 A schematic diagram of an AI digital human construction method based on 3D modeling technology in an embodiment of the present invention;

[0071] Figure 2 Schematic diagram of an AI digital human construction system based on 3D modeling technology in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0073] The embodiment of the present invention provides an AI digital human construction method based on 3D modeling technology, such as Figure 1 As shown, including:

[0074] Step 1: Obtain the digital human construction requirement basis information of the vending machine, which includes: character image selection instructions and captured images; wherein the character image selection instructions are: the digital human construction instructions input by the user of the vending machine (issued by the user by clicking the command trigger button preset on the screen of the vending machine); the captured images are: the captured images of the user of the vending machine (captured by the camera preset on the vending machine);

[0075] Step 2: Determine the digital human construction requirements of the vending machine based on the digital human construction requirements; when determining the digital human construction requirements of the vending machine, first obtain the digital human construction instructions. If the acquisition is successful, directly parse the digital human construction instructions to determine. If the digital human construction instructions are not identified, that is, when the digital human construction requirements are based on the captured image, analyze the user characteristics to predict the requirements. When predicting the requirements, if the user's historical construction records are identified based on the captured image, directly determine the construction requirements based on the historical construction records. Otherwise, filter the construction requirements with the same user profile based on the user profile.

[0076] Step 3: Design the digital human construction according to the digital human construction requirements and obtain the design plan; the design plan is: retrieve the plan of digital human construction materials from the material library (for example: 3D modeling materials with brown hair and blue eyes);

[0077] Step 4: Based on 3D modeling technology, construct AI digital human according to the design plan.

[0078] The working principle and beneficial effects of the above technical solution are:

[0079] The present invention obtains the digital human construction demand basis information of the vending machine, and the digital human construction demand basis information includes the character image selection instruction actively input by the user and the photographed image passively taken. Based on different digital human construction demand basis information, the digital human construction demand is determined, and the demand determination priority of the character image selection instruction is higher than the photographed image. When the user uses the vending machine for the first time, after the character image selection instruction is triggered to construct the digital character image, the system records the user's construction record. When the user uses it again next time, unless the user actively triggers the character image selection instruction again, the system will give priority to reading the historical construction record to determine the construction demand, and there is no need to set it every time. If the user has neither triggered the character image selection instruction nor has a construction record in history, the construction demand of the same user portrait is adaptively obtained; based on the digital human construction demand, the digital human construction design is carried out to obtain the design plan, and the AI ​​digital human is constructed based on the 3D modeling technology and the design plan. The customized digital human can communicate with the user, which is more humane and improves the user experience of the vending machine.

[0080] In one embodiment, obtaining the digital human construction requirement information of the vending machine includes:

[0081] Acquire a character image selection instruction of a user who uses a vending machine to purchase a commodity; wherein acquiring the character image selection instruction comprises: analyzing a selection signal of a character image selection item displayed on a screen of the vending machine by the user or a voice of the user selecting a character image, and determining the character image selection instruction;

[0082] If the acquisition is successful, the character image selection instruction is used as the basis information for the digital human construction requirements;

[0083] If the acquisition fails, an image of the user who uses the vending machine to purchase goods is obtained by a preset camera device of the vending machine, and the image is used as the basis information for the digital human construction requirements.

[0084] The working principle and beneficial effects of the above technical solution are:

[0085] The present invention first obtains the character image selection instruction actively input by the user who uses the vending machine to buy goods. If the acquisition is successful, the digital human construction demand basis information of the digital human construction demand can be directly determined. Otherwise, based on the photographed image of the user who buys goods at the vending machine, the digital human construction demand basis information is predicted. Using two methods to obtain the digital human construction demand basis information is more comprehensive.

[0086] In one embodiment, according to the digital human construction requirement basis information, determining the digital human construction requirement of the vending machine includes:

[0087] Acquiring the information type of the digital human construction demand basis information; wherein the information type is: the basis type of the digital human construction demand basis information, such as: system triggering instructions, device shooting images, etc.;

[0088] When the information type is a character image selection instruction, the character image selection instruction is parsed to obtain the digital human construction requirements;

[0089] When the information type is captured images, the digital human construction requirements are predicted based on the captured images.

[0090] The working principle and beneficial effects of the above technical solution are:

[0091] The present invention obtains the information type of the digital human construction demand based on the information; when the information type is a character image selection instruction, the character image selection instruction is parsed to obtain the digital human construction demand; when the information type is a captured image, the digital human construction demand is predicted based on the captured image, and the digital human construction demand is determined in combination with the specific purchase situation, thereby improving the rationality of obtaining the digital human construction demand.

[0092] In one embodiment, predicting the digital human construction requirements based on the captured images includes:

[0093] According to the captured image, the historical construction record of the user is obtained; wherein the historical construction record is: the record of the user inputting instructions on the vending machine to construct the digital human in the past;

[0094] If the acquisition is successful, the digital human construction requirements are determined based on the historical construction records; when the digital human construction requirements are determined based on the historical construction records, the digital human image constructed in the historical construction records is directly used as the digital human construction requirements;

[0095] If the acquisition fails, based on the preset character portrait extraction template, the character portrait is extracted according to the captured image; wherein the preset character portrait extraction template is: a predefined framework, which includes multiple feature extraction rules, and the feature extraction rules are used to identify and extract different features and attributes of the character from the captured image, such as: facial feature extraction rules, clothing feature extraction rules; the character portrait is: a character image description vector constructed based on the character features, such as: [old people, wearing low color saturation], and another example: [young people, wearing low color saturation];

[0096] Determine a historical figure portrait demand probability table according to the historical figure image selection instruction; wherein the historical figure portrait demand probability table includes: historical figure portraits and selection probabilities corresponding to different historical figure image selection instructions, and obtain the historical figure image selection instructions and corresponding selected historical figure portraits according to the historical figure image selection instructions recorded in the triggering records of the historical figure image selection instructions;

[0097] According to the demand probability table of character portraits and historical character portraits, determine the digital human model with the highest demand probability corresponding to the historical character portrait that matches the person-household portrait;

[0098] Use the digital human model as the digital human construction requirement.

[0099] The working principle and beneficial effects of the above technical solution are:

[0100] When predicting the digital human construction requirements based on the captured images, the system will give priority to reading the historical construction records to determine the construction requirements. If there is no construction record in history, the character portrait is extracted based on the introduced character portrait extraction template and the captured image. The trigger records of the historical character image selection instructions are used to calculate the historical character portrait requirement probability table. Based on the character portrait and the historical character portrait requirement probability table, the digital human model with the highest probability of corresponding to the historical character portrait that matches the character portrait is determined, and the digital human model is used as the digital human construction requirement. Based on the real-time acquired character image and the historical character portrait requirement probability table that has been calculated historically, the digital human model with the highest probability of corresponding to the character image is determined as the digital human construction requirement, so that the acquisition of digital human construction requirements is more reasonable.

[0101] In one embodiment, the digital human construction design is performed according to the digital human construction requirements, and the design scheme is obtained, including:

[0102] Based on the preset demand feature extraction template, according to the digital human construction requirements, the digital human construction demand features are extracted; wherein the preset demand feature extraction template is: the description features of the digital human construction requirements, such as: part type and part description;

[0103] According to the demand characteristics and the preset digital human material library, a design solution is obtained. The preset digital human material library includes: one-to-one correspondence between pre-selected demand characteristics and digital human materials, and the pre-selected demand characteristics are descriptive characteristics of the corresponding digital human materials.

[0104] The working principle and beneficial effects of the above technical solution are:

[0105] The present invention introduces a demand feature extraction template, extracts the demand features of digital human construction according to the digital human construction requirements, introduces a digital human material library, determines the design scheme corresponding to the demand features, and improves the suitability of the design scheme.

[0106] The embodiment of the present invention provides an AI digital human construction method based on 3D modeling technology, which also includes:

[0107] Supplement the digital human material library;

[0108] Among them, the digital human material library is supplemented, including:

[0109] Based on the digital human resource big data platform, the digital human material library is supplemented; the digital human resource big data platform is: a big data platform that provides digital human shared resources;

[0110] and / or,

[0111] The digital human material library is supplemented based on the uploaded materials of the local supplementary nodes. Among them, the local supplementary nodes are: the network nodes of the 3D modeling personnel of the software service provider of the vending machine.

[0112] The working principle and beneficial effects of the above technical solution are:

[0113] The present invention introduces two ways to supplement the digital human material library. The first way is to connect to the digital human resource big data platform to supplement the digital human material library. The second way is to connect to the network node of the three-dimensional modeling personnel of the software service provider of the vending machine to supplement the digital human material library, thereby increasing the richness of the digital human material library and improving the user experience in subsequent applications.

[0114] In one embodiment, based on the digital human resource big data platform, the digital human material library is supplemented, including:

[0115] Based on the Digital Human Resource Big Data Platform, the shared information of shared materials is obtained; the shared materials are: the Digital Human construction materials shared in the Digital Human Resource Big Data Platform; the shared information is: the number of times the shared materials have been viewed, the number of times they have been downloaded, the download requirements, and user evaluations;

[0116] Extracting shared information features based on a preset shared information feature extraction template and based on the shared information; wherein the preset shared information feature extraction template is a predefined template for extracting shared information features (information content) of different types of shared information, such as identifying the number of times a shared material has been viewed from the shared information, identifying the download requirements of the shared material from the shared information, etc.; shared information features include: the number of times a shared material has been viewed, the number of downloads, the download requirements and user evaluations, etc.;

[0117] According to the shared information type of the shared information feature, the shared material heat quantification index type is determined; the shared material heat quantification index type includes: weight index and numerical index; wherein, the weight index is: there is a correlation between the presence or absence of the corresponding shared information feature and the size of the shared material heat quantification value, for example: if the download requirement is no, then the weight index value is: 0.8, if the download requirement is yes, then the shared material heat quantification index value (weight index value) is: 1.2; the numerical index is: the size of the shared material heat quantification value directly characterized by the corresponding shared information feature, for example: if the number of downloads is 10, then the shared material heat quantification index value (numerical index value) is: 10, if the number of downloads is 50, then the shared material heat quantification index value (numerical index value) is: 50; the determination logic of the shared material heat quantification index type according to the shared information type is preset manually;

[0118] Determine the shared material heat quantification index value according to different shared material heat quantification index types; wherein the conversion rules for converting different shared information features into the shared material heat quantification index values ​​of the corresponding shared material heat quantification index types are manually set based on experience;

[0119] Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index type is a weight index to obtain a target weight value;

[0120] Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index types are numerical indicators to obtain a target sum value;

[0121] Multiply the target weight value and the target sum value to obtain the thermal value of the shared material;

[0122] Compare the material similarity between the shared material with the largest heat value and the local real-time material in the digital human material library. If the material similarity is less than the preset material similarity threshold, then connect the shared path to supplement the digital human material library according to the sharing information of the corresponding shared material. Among them, the material similarity is: the material similarity between the shared material and the local real-time material in the digital human material library; the preset material similarity threshold is manually pre-set, such as: 0.9; the shared path is: the shared network path of the shared material with the largest shared heat value and the material similarity is less than the preset material similarity threshold.

[0123] The working principle and beneficial effects of the above technical solution are:

[0124] The present invention obtains the shared information of the shared materials of the digital human resource big data platform, extracts the shared information features of the shared information based on the shared information feature extraction template, determines the shared material heat quantification index type and the shared material heat quantification index value according to the shared information type of the shared information feature, cumulatively calculates the shared material heat quantification index values ​​of the shared material heat quantification index type as a weight index to obtain the target weight value, and cumulatively calculates the shared material heat quantification index values ​​of the shared material heat quantification index type as a numerical index to obtain the target sum value, multiplies the target weight value and the target sum value correspondingly to obtain the thermal value of the shared material, thereby improving the accuracy of the thermal value calculation; the shared materials with the highest heat are screened in real time for comparison with the local real-time materials, and the shared materials with high heat that do not exist in the digital human material library are supplemented, while avoiding missing the high-heat materials, duplication with the local materials is also avoided, thereby improving the efficiency of material supplementation.

[0125] In one embodiment, the digital human material library is supplemented based on the uploaded material of the local supplement node, including:

[0126] Obtaining user evaluation records of the vending machine; wherein the user evaluation record is: a record of the user's evaluation of the service after the vending machine is used;

[0127] Parse the user evaluation record to obtain the user evaluation semantics; wherein the user evaluation semantics is: the semantics obtained by analyzing the user evaluation record based on the semantic analysis technology;

[0128] Determine the material evaluation record in the user evaluation record according to the user evaluation semantics and the preset material description semantics; wherein the preset material description semantics is: a language expression describing the attributes and characteristics of the digital human material, which is preset manually;

[0129] Expand the material evaluation records on a preset time axis based on the generation time, and obtain the time density of the material evaluation records within the preset time interval; wherein the preset time interval is: every month; the time density is: the number of material evaluation records within a month;

[0130] If the time density is greater than or equal to the preset time density threshold, the corresponding time interval is used as the target time interval; wherein the preset time density threshold is manually preset, for example: 5 per month;

[0131] Obtaining material evaluation features of each material evaluation record within a target time interval; wherein the material evaluation features are: material evaluation semantics;

[0132] Performing feature clustering on the material evaluation features; wherein, performing feature clustering on the material evaluation features includes: performing semantic clustering on similar material evaluation semantics;

[0133] If clustering is successful, obtain the feature clustering result; wherein the feature clustering result is: similar material evaluation semantic set;

[0134] Based on the feature clustering results, determine the material supplement demand; wherein the material supplement demand is: the demand information for supplementing the digital human material;

[0135] Send the material supplementation requirements to the local supplementation node, and obtain the uploaded materials corresponding to the material supplementation requirements uploaded by the local supplementation node to supplement the digital human material library. Among them, the local supplementation node is: the network node of the 3D modeler of the software service provider of the vending machine. After receiving the material supplementation requirements, the software service provider corresponding to the local supplementation node will perform manual modeling based on the material supplementation requirements and add the modeling results to the digital human material library.

[0136] The working principle and beneficial effects of the above technical solution are:

[0137] The present invention obtains user evaluation semantics based on the user evaluation records of the vending machine, performs semantic matching between the user evaluation semantics and the preset material description semantics, and if there is a semantic match, uses the corresponding user evaluation record as the material evaluation record; expands the material evaluation record on the preset time axis based on the generation time, and obtains the time density of the material evaluation record in the time interval; screens the target time interval with a time density greater than the time density threshold, and extracts the material evaluation features of the material evaluation records in the target time interval, performs feature clustering on the material evaluation features to obtain feature clustering results, determines the material supplement demand based on the feature clustering results, sends the material supplement demand to the local supplement node, and obtains the uploaded material supplement digital human material library corresponding to the material supplement demand uploaded by the local supplement node. When similar material evaluations appear in the service evaluations fed back by users, the present invention promptly sends the material supplement demand to the local supplement node corresponding to the modeler, and the local supplement node promptly uploads the corresponding material, thereby improving the service quality.

[0138] In one embodiment, clustering is successful, including:

[0139] The target number of cluster members in the feature cluster is greater than the number of first evaluation records corresponding to the time density; wherein the target number of members is the number of cluster members in the feature cluster for determining whether the clustering success condition is met, such as the number of similar evaluation semantics; the first evaluation record number is the number of material evaluation records within the target time interval;

[0140] or,

[0141] The target number of cluster members in the feature cluster is greater than the number of cluster members in other feature clusters, and the target number of members is less than the first number of evaluation records corresponding to the time density and greater than the second number of evaluation records corresponding to the time density threshold. The number of cluster members in other feature clusters is: the number of cluster members in the cluster cluster after clustering other cluster members, except for the cluster members in the feature clusters that determine whether the clustering success condition is met; the second number of evaluation records is: the number of material evaluation records in the target time interval when the time density in the target time interval is equal to the time density threshold.

[0142] The working principle and beneficial effects of the above technical solution are:

[0143] The present invention introduces two conditions to limit clustering success. The first one is that the target number of cluster members is greater than the number of evaluation records in the corresponding target time interval, indicating that the corresponding evaluation features appear densely and more than once in each record, and the clustering is directly determined to be successful. The second one is that the target number of cluster members is less than the number of first evaluation records and greater than the corresponding second evaluation records, indicating that although the second evaluation record number does not appear repeatedly in a material evaluation record, the total number of occurrences reaches the specified value, and it is the largest number of cluster members in all cluster clusters, and the clustering is determined to be successful. The embodiment of the present invention introduces a clustering success condition to limit the characteristic clustering results, screens the basis for generating material supplement requirements, and improves the efficiency of generating subsequent material supplement requirements.

[0144] The embodiment of the present invention provides an AI digital human construction system based on 3D modeling technology, such as Figure 2 As shown, including:

[0145] According to the information acquisition subsystem 1, it is used to obtain the digital human construction requirement basis information of the vending machine, and the digital human construction requirement basis information includes: character image selection instructions and shot images;

[0146] A digital human construction requirement determination subsystem 2, used for determining the digital human construction requirement of the vending machine according to the digital human construction requirement basis information;

[0147] Design scheme acquisition subsystem 3, used to design the digital human construction according to the digital human construction requirements and obtain the design scheme;

[0148] The digital human construction subsystem 4 is used to construct an AI digital human based on the 3D modeling technology and according to the design plan.

[0149] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The AI ​​digital human construction method based on 3D modeling technology is characterized by: include: Obtaining the digital human construction requirement basis information of the vending machine, the digital human construction requirement basis information including: character image selection instructions and shot images; Determine the digital human construction requirements for the vending machine based on the digital human construction requirements information; Design the digital human construction according to the digital human construction requirements and obtain the design plan; Based on 3D modeling technology, AI digital humans are constructed according to the design plan.

2. The AI ​​digital human construction method based on 3D modeling technology according to claim 1, characterized in that: Obtain the information required to build the digital human of the vending machine, including: Obtaining a character image selection instruction of a user who uses a vending machine to purchase goods; If the acquisition is successful, the character image selection instruction is used as the basis information for the digital human construction requirements; If the acquisition fails, an image of the user who uses the vending machine to purchase goods is obtained by a preset camera device of the vending machine, and the image is used as the basis information for the digital human construction requirements.

3. The AI ​​digital human construction method based on 3D modeling technology according to claim 1, characterized in that: According to the digital human construction requirements information, the digital human construction requirements of the vending machine are determined, including: The type of information on which the digital human construction needs to be based; When the information type is a character image selection instruction, the character image selection instruction is parsed to obtain the digital human construction requirements; When the information type is captured images, the digital human construction requirements are predicted based on the captured images.

4. The AI ​​digital human construction method based on 3D modeling technology as claimed in claim 3, characterized in that: Based on the captured images, predict the requirements for building digital humans, including: Obtain the user's historical construction records based on the captured images; If the acquisition is successful, the digital human construction requirements are determined based on the historical construction records; If the acquisition fails, based on the preset character portrait extraction template, the character portrait is extracted according to the captured image; According to the historical figure image selection instructions, determine the historical figure portrait demand probability table; According to the demand probability table of character portraits and historical character portraits, determine the digital human model with the highest demand probability corresponding to the historical character portrait that matches the person-household portrait; Use the digital human model as the digital human construction requirement.

5. The AI ​​digital human construction method based on 3D modeling technology according to claim 1, characterized in that: Design the digital human construction according to the digital human construction requirements and obtain the design plan, including: Based on the preset demand feature extraction template, the demand features of digital human construction are extracted according to the digital human construction requirements; Obtain design solutions based on required features and the preset digital human library.

6. The AI ​​digital human construction method based on 3D modeling technology as claimed in claim 5, characterized in that: Also includes: Supplement the digital human material library; Among them, the digital human material library is supplemented, including: Based on the digital human resource big data platform, supplement the digital human material library; and / or, Upload materials based on local supplementary nodes to supplement the digital human material library.

7. The AI ​​digital human construction method based on 3D modeling technology according to claim 6, characterized in that: Based on the digital human resource big data platform, the digital human material library is supplemented, including: Based on the digital human resource big data platform, obtain the shared information of shared materials; Extracting shared information features based on a preset shared information feature extraction template and shared information; Determine the type of shared material heat quantification index according to the shared information characteristics; the shared material heat quantification index type includes: weight index and value index; Determine the shared material heat quantification index value according to different shared material heat quantification index types; Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index type is a weight index to obtain a target weight value; Accumulate and calculate the shared material heat quantization index values ​​whose shared material heat quantization index types are numerical indicators to obtain a target sum value; Multiply the target weight value and the target sum value to obtain the thermal value of the shared material; Compare the material similarity between the shared material with the largest heat value and the local real-time material of the digital human material library. If the material similarity is less than the preset material similarity threshold, the digital human material library is supplemented by connecting the sharing path according to the sharing information of the corresponding shared material.

8. The AI ​​digital human construction method based on 3D modeling technology as claimed in claim 6, characterized in that: Upload materials based on local supplementary nodes to supplement the digital human material library, including: Get the user evaluation record of the vending machine; Parse user evaluation records and obtain user evaluation semantics; Determine the material evaluation record in the user evaluation record according to the user evaluation semantics and the preset material description semantics; Expand the material evaluation records on a preset time axis based on the generation time, and obtain the time density of the material evaluation records within a preset time interval; If the time density is greater than or equal to the preset time density threshold, the corresponding time interval is used as the target time interval; Obtaining material evaluation features of each material evaluation record within a target time interval; Perform feature clustering on material evaluation features; If clustering is successful, obtain the feature clustering results; Determine the need for additional materials based on feature clustering results; The material supplementation requirement is sent to the local supplementation node, and the uploaded material corresponding to the material supplementation requirement uploaded by the local supplementation node is obtained to supplement the digital human material library.

9. The AI ​​digital human construction method based on 3D modeling technology as claimed in claim 8, characterized in that: Clustering success includes: The target number of cluster members in the feature clustering is greater than the number of first evaluation records corresponding to the time density; or, The target number of cluster members in the feature cluster is greater than the number of cluster members in other feature clusters, and the target number of members is less than the first number of evaluation records corresponding to the time density and greater than the second number of evaluation records corresponding to the time density threshold.

10. The AI ​​digital human construction system based on 3D modeling technology is characterized by: include: The information acquisition subsystem is used to acquire the digital human construction requirement basis information of the vending machine, and the digital human construction requirement basis information includes: character image selection instructions and shot images; A digital human construction requirement determination subsystem is used to determine the digital human construction requirement of the vending machine according to the digital human construction requirement basis information; The design scheme acquisition subsystem is used to design the digital human construction according to the digital human construction requirements and obtain the design scheme; The digital human construction subsystem is used to construct AI digital humans based on 3D modeling technology and according to the design plan.

Images