An e-commerce platform system that uses virtual human technology to realize virtual trial of beauty products

Abstract

The present invention discloses an e-commerce platform system for realizing virtual trial of beauty products by using virtual human technology, which relates to the field of virtual trial, including: a login module confirms the identity information of consumers and e-commerce companies, and performs login management; an information acquisition module acquires consumer information and beauty product information; a virtual module generates a corresponding virtual human model according to consumer information, generates a corresponding virtual beauty product according to beauty product information, applies the virtual beauty product selected by the consumer to the virtual human model according to the interactive information of the consumer received by the interactive module, obtains a virtual trial model, and displays it through the interactive module; a storage module stores and updates identity information, as well as consumer information, beauty product information, virtual human model, virtual beauty product, and virtual trial model corresponding to the identity information. The present invention can intuitively display the effect of consumers using beauty products, thereby improving the shopping experience of the e-commerce platform.

Description

Technical Field

[0001] The present invention relates to the field of virtual trial technology, and more specifically to an e-commerce platform system for realizing virtual trial of beauty products using virtual human technology. Background Art

[0002] With the development of Internet technology, online shopping is increasing day by day. When shopping online, people can get a general understanding of the products through pictures and texts provided by merchants, or in the form of videos.

[0003] For items with fixed shapes, pictures, texts, and videos can meet consumers' needs for understanding products. However, for products that consumers need to actually wear or try out, especially beauty products such as foundation, lipstick, and eye shadow, it is often difficult for consumers to intuitively understand the effects of using beauty products, resulting in refunds and returns of inappropriate products, which not only leads to poor shopping experience for users, but also increases e-commerce logistics costs.

[0004] Therefore, how to intuitively display the effects of beauty products used by consumers, thereby improving the shopping experience on e-commerce platforms, is an issue that technical personnel in this field urgently need to solve. Summary of the invention

[0005] In view of this, the present invention provides an e-commerce platform system that uses virtual human technology to achieve virtual trial of beauty products, which can perform virtual trial of beauty products, demonstrate the effects of consumers using beauty products, and improve shopping experience.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] The present invention discloses an e-commerce platform system for realizing virtual trial of beauty products by using virtual human technology, comprising: a login module, an information acquisition module, a virtual module, an interaction module, and a storage module;

[0008] The login module confirms the identity information of the consumer and the e-commerce company and performs login management;

[0009] The information acquisition module is used to acquire consumer information and beauty product information;

[0010] The virtual module generates a corresponding virtual human model according to the consumer information; generates a corresponding virtual beauty product according to the beauty product information; and applies the virtual beauty product selected by the consumer to the virtual human model according to the consumer interaction information received by the interaction module, obtains a virtual trial model, and displays it through the interaction module;

[0011] The storage module is used to store and update the identity information, as well as the consumer information corresponding to the identity information, the beauty product information, the virtual human model, the virtual beauty product, and the virtual trial model.

[0012] Further, the login module includes: a consumer login unit and an e-commerce login unit, which are used for identity confirmation and login management of consumers and e-commerce, respectively;

[0013] The information acquisition module includes: a consumer information acquisition unit and a beauty product information acquisition unit, which respectively acquire the consumer information and the beauty product information;

[0014] The virtual module includes a virtual human unit, a beauty product unit, and a virtual trial unit, which are respectively used to generate the virtual human model, the virtual beauty product, and the virtual trial model.

[0015] Furthermore, the consumer information includes consumer head picture data and facial key point data;

[0016] The head picture data is a two-dimensional picture including the head of the consumer taken at multiple different angles;

[0017] The facial key point data is obtained by moving the key points of the standard face model to determine the corresponding consumer face model after the consumer sets the length, width and height of the standard face model through the interaction module according to the consumer's own facial features, and then the information acquisition module reads the three-dimensional coordinates of the standard face model to obtain the facial key point data;

[0018] The cosmetic products are divided into wearable cosmetic products and application cosmetic products;

[0019] The beauty product information of the wearable beauty product includes: image data, three-dimensional scanning data, transparency, surface reflectivity, light scattering, and roughness;

[0020] The beauty product information of the coating beauty product includes: color, color saturation, transparency, reflectivity, light scattering, and roughness.

[0021] Furthermore, the generating of the corresponding virtual human model includes:

[0022] Determine the shooting distance of each picture according to the shooting focal length of each picture;

[0023] Extracting feature points from the head image data to determine facial key points contained in each image;

[0024] Based on the shooting distance of each picture, the two-dimensional plane coordinates of the facial key points, and the three-dimensional coordinates of the facial key point data, the three-dimensional posture of the shooting device relative to the consumer face model when shooting each picture is solved to obtain a virtual camera posture set;

[0025] Selecting two 3D postures that meet the set virtual camera spacing threshold and angle threshold from the virtual camera posture set as a group, performing binocular vision 3D reconstruction based on the corresponding two images, and obtaining a partial head 3D model of the consumer;

[0026] Selecting multiple groups of three-dimensional postures, performing binocular vision three-dimensional reconstruction, and splicing the facial key point coordinates in the partial head three-dimensional model obtained by reconstruction to obtain a complete head three-dimensional model;

[0027] The three-dimensional head model is subjected to surface smoothing processing to obtain the virtual human model.

[0028] Furthermore, before the feature point extraction, the head image data is preprocessed, including: firstly detecting the edge contour of the human body; then removing the background pixels according to the edge contour of the human body; and then converting the removed image from the RGB space to the HSV space.

[0029] Furthermore, the generating of the corresponding virtual beauty products includes:

[0030] For the wearable cosmetic product, determining a three-dimensional model of the wearable cosmetic product according to the three-dimensional scanning data, identifying a contour line of the three-dimensional model, and then determining a plurality of surfaces of the three-dimensional model according to the contour line;

[0031] Performing contour recognition on the image data to determine the contour line of the wearable cosmetic product, and determining a plurality of surfaces according to the recognized contour line;

[0032] Determining surface mapping data of the three-dimensional model according to a correspondence between the three-dimensional model and a surface determined by the image data;

[0033] Determine the surface material data of the three-dimensional model according to the transparency, surface reflectivity, light scattering, and roughness of the wearable cosmetic product to obtain a wearable virtual cosmetic product;

[0034] For the coated beauty products, surface color data is directly generated according to the color and color saturation of the coated beauty products, and surface material data is generated according to the transparency, reflectivity, light scattering, and roughness of the coated beauty products. The surface color data and surface material data are used as coated virtual beauty products.

[0035] Furthermore, obtaining the virtual trial model includes:

[0036] For the wearable cosmetics product, the consumer directly splices the three-dimensional model of the wearable virtual cosmetics product onto the virtual human model through the interactive module to obtain the virtual trial model;

[0037] For the coating type beauty products, the consumer first selects the coating area and coating amount on the virtual human model, then determines the color of the coated area after coating according to the coating color model, and uses the surface material data of the coating type virtual beauty products as the surface material of the coating area. Finally, the coating area selected on the virtual human model is modified according to the determined surface color and surface material to obtain the virtual trial model.

[0038] Furthermore, the coating color model is a three-layer fully connected neural network, the number of hidden layers in the three-layer neural network is 8, 8, and 1 respectively, the fully connected neural network uses the relu function as an activation function, and the loss function is the mean square error.

[0039] Furthermore, the coating color model is trained in the following manner:

[0040] Obtain beauty products with different beauty product information, conduct coating experiments with different coating thicknesses on skin-like materials with different base colors, and collect the final coated area’s post-coating color;

[0041] The color data, color saturation data, transparency data, reflectivity data, light scattering data, roughness data, coating thickness data, and coating base color data of the experimental cosmetics product are used as sample features, and the color data after coating is used as a sample label to form a sample data set, and the multiple sample data sets constitute a first sample data set;

[0042] The first sample data set is used to train a fully connected neural network to obtain a universal coating color model.

[0043] Furthermore, it also includes that the e-commerce company conducts the coating experiment on the beauty products it sells, collects the coated color of the final coated area, and organizes it to obtain a second sample data set, and uses the second sample data set to adjust and train the universal coating color model to obtain the coating color model of each e-commerce product; and uses the Adam optimizer and 5-fold cross validation to perform the adjustment training.

[0044] Through the above technical solutions, it can be known that compared with the prior art, the present invention discloses an e-commerce platform system that uses virtual human technology to realize virtual trial of beauty products. Through feature point extraction and camera three-dimensional posture inverse solution, it can take pictures multiple times when consumers use a single shooting device, and build a three-dimensional model based on binocular vision, so as to accurately reconstruct the three-dimensional model of the consumer's head, and ensure that the effect of virtual makeup trial is more realistic and credible. By establishing a beauty product model and cooperating with the established virtual human model, consumers can adjust the trial models of different products in real time, and instantly see the use effects of different beauty products, realize instant feedback and interaction, and improve user participation; and using a coating color model based on deep learning, it can more accurately present the effect of beauty products according to the consumer's skin color and coating amount. The present invention realizes the virtual trial of beauty products through virtual human technology, and can intuitively display the effect of consumers using beauty products, thereby improving the shopping experience of the e-commerce platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0046] Figure 1 It is a schematic diagram of the overall structure of an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The embodiment of the present invention discloses an e-commerce platform system for realizing virtual trial of beauty products by using virtual human technology, such as Figure 1 As shown, it includes: a login module, an information acquisition module, a virtual module, an interaction module, and a storage module;

[0049] The login module confirms the identity information of consumers and e-commerce companies and performs login management;

[0050] The information acquisition module is used to obtain consumer information and beauty product information;

[0051] The virtual module generates a corresponding virtual human model according to the consumer information; generates a corresponding virtual beauty product according to the beauty product information; applies the virtual beauty product selected by the consumer to the virtual human model according to the consumer's interactive information received by the interactive module, obtains a virtual trial model, and displays it through the interactive module;

[0052] The storage module is used to store and update identity information, as well as consumer information, beauty product information, virtual human models, virtual beauty products, and virtual trial models corresponding to the identity information.

[0053] In a specific embodiment, the login module includes: a consumer login unit and an e-commerce login unit, which are used for identity confirmation and login management of consumers and e-commerce respectively; login management can be performed using passwords, fingerprints, face recognition, etc.

[0054] The information acquisition module includes: a consumer information acquisition unit and a beauty product information acquisition unit, which respectively acquire consumer information and beauty product information; consumers or e-commerce companies control cameras, three-dimensional scanners, etc. through the interactive module to acquire images or color information, and can also directly fill in facial feature dimensions, three-dimensional data of beauty products, transparency, reflectivity, light scattering, roughness, etc. through the interactive module (such as the mobile phone or tablet app interactive interface, or computer screen keyboard and mouse operations).

[0055] The virtual module includes a virtual human unit, a beauty product unit, and a virtual trial unit, which are used to generate a virtual human model, a virtual beauty product, and a virtual trial model, respectively.

[0056] In a specific embodiment, the consumer information includes consumer head picture data and facial key point data;

[0057] The head image data is a two-dimensional image of the consumer's head taken at multiple different angles;

[0058] The facial key point data is obtained by setting the length, width and height of the standard face model through the interactive module according to the consumer's own facial features, and then moving the key points of the standard face model to determine the corresponding consumer face model, and then the information acquisition module reads the three-dimensional coordinates of the standard face model to obtain the facial key point data. Among them, there are multiple different standard face models according to the proportion and size of the face, which are matched according to the size set by the consumer or selected by the consumer; after selection, the consumer adjusts the position of the key points on the standard face model according to his own facial features, so that the standard model is close to the consumer's own facial features. The key points on the standard face model can be determined according to the universal facial 21-point, 35-point, and 68-point marking serial number standards.

[0059] Beauty products are divided into wearable beauty products, such as eyelashes, earrings and hair accessories, and applied beauty products, such as lipstick, eye shadow, foundation, etc.

[0060] The beauty product information of wearable beauty products includes: image data, 3D scanning data, transparency, surface reflectivity, light scattering, and roughness;

[0061] The beauty product information of coated beauty products includes: color, color saturation, transparency, reflectivity, light scattering, and roughness.

[0062] In a specific embodiment, generating a corresponding virtual human model includes:

[0063] Determine the shooting distance of each picture according to the shooting focal length of each picture;

[0064] Extract feature points from head image data to determine facial key points contained in each image;

[0065] Based on the shooting distance of each picture, the two-dimensional plane coordinates of the facial key points, and the three-dimensional coordinates of the facial key point data, the three-dimensional posture of the shooting device relative to the consumer's face model when shooting each picture is solved to obtain a virtual camera posture set. According to the shooting distance and the 2D-3D coordinates of the shooting key points, the three-dimensional posture of the camera is reversely solved, usually using multiple pairs of matching 2D-3D points (usually more than four pairs) for matrix solution, so as to describe the relative posture between the camera coordinate system and the face model coordinate system. The specific solution process belongs to the existing technology and will not be repeated here.

[0066] Two three-dimensional postures that meet the set virtual camera spacing threshold and angle threshold are selected as a group in the virtual camera posture set, and binocular vision three-dimensional reconstruction is performed based on the corresponding two pictures to obtain a partial three-dimensional model of the consumer's head; the camera spacing and angle will have a greater impact on the field of view. When it is larger, it will make it difficult to align the overlapping areas, and when it is smaller, it will cause the depth estimation to become inaccurate, resulting in reduced two-dimensional reconstruction accuracy. Therefore, consumers are required to take as many pictures as possible when taking pictures to improve the three-dimensional reconstruction accuracy.

[0067] Select multiple groups of 3D postures, perform binocular vision 3D reconstruction, and splice the facial key point coordinates in the reconstructed partial head 3D model to obtain a complete head 3D model;

[0068] The three-dimensional head model is subjected to surface smoothing, for example, by Gaussian smoothing, vertex offset smoothing, etc., to obtain a virtual human model.

[0069] In a specific embodiment, before extracting feature points, the head image data is also preprocessed, including: firstly detecting the edge contour of the human body; then removing background pixels according to the edge contour of the human body; and then converting the removed image from the RGB space to the HSV space.

[0070] In a specific embodiment, generating corresponding virtual beauty products includes:

[0071] For wearable cosmetics, determine the 3D model of the wearable cosmetics based on the 3D scanning data, identify the contour of the 3D model, and then determine several surfaces of the 3D model based on the contour;

[0072] Contour recognition is performed on the image data to determine the contour lines of wearable beauty products, and several surfaces are determined based on the identified contour lines. After contour recognition is performed on the image data, the distortion of the contour lines can also be corrected to facilitate better surface relationship matching.

[0073] Determine surface mapping data of the three-dimensional model according to the correspondence between the three-dimensional model and the surface determined by the image data;

[0074] According to the transparency, surface reflectivity, light scattering and roughness of the wearable cosmetics, the surface material data of the three-dimensional model is determined to obtain the wearable virtual cosmetics.

[0075] For coated beauty products, surface color data is directly generated based on the color and color saturation of the coated beauty products. Surface material data is generated based on the transparency, reflectivity, light scattering, and roughness of the coated beauty products. The surface color data and surface material data are used as coated virtual beauty products.

[0076] In a specific embodiment, obtaining a virtual trial model includes:

[0077] For wearable beauty products, consumers can directly splice the 3D model of the wearable virtual beauty product onto the virtual human model through the interactive module to obtain a virtual trial model;

[0078] For cosmetic products that are coated, consumers first select the coating area and coating amount on the virtual human model, then determine the color of the coated area based on the coating color model, and use the surface material data of the virtual cosmetic product as the surface material of the coated area. Finally, modify the selected coating area on the virtual human model based on the determined surface color and surface material to obtain a virtual trial model. After obtaining the virtual model, three-dimensional rendering is performed based on the virtual model's three-dimensional data, surface mapping, surface color, surface material data, and the viewing angle and lighting angle selected by the customer, and displayed directly to consumers through the interactive module.

[0079] In a specific embodiment, the coating color model is a three-layer fully connected neural network, the number of hidden layers in the three-layer neural network is 8, 8, and 1 respectively, the fully connected neural network uses the relu function as the activation function, and the loss function is the mean square error.

[0080] In a specific embodiment, the coating color model is trained as follows:

[0081] Obtain beauty products with different beauty product information, conduct coating experiments with different coating thicknesses on skin-like materials with different base colors, and collect the final coated area’s post-coating color;

[0082] The color data, color saturation data, transparency data, reflectivity data, light scattering data, roughness data, coating thickness data, and coating base color data of the experimental cosmetics product are used as sample features, and the color data after coating is used as a sample label to form a sample data set, and the multiple sample data sets constitute a first sample data set;

[0083] The first sample data set is used to train a fully connected neural network to obtain a universal coating color model.

[0084] In a specific embodiment, it also includes that the e-commerce company conducts a coating experiment on the beauty products it sells, collects the coated color of the final coated area, and then sorts out a second sample data set, uses the second sample data set to adjust and train the universal coating color model to obtain the coating color model of each e-commerce product; and uses the adam optimizer and 5-fold cross validation for adjustment and training. The adjustment and training process includes model loading, fine-tuning and model storage. During the fine-tuning process, the adam optimizer and 5-fold random cross validation are used, the training batch size is 64, and the round is 1000. Due to the introduction of the second sample data set, adjustment and training are performed for different e-commerce beauty products, and the weights of each hidden layer of the fully connected neural network will be fine-tuned, so that the universal model can be targeted for the beauty products of different e-commerce companies. The model can be more accurately predicted. The coating color. It should be understood that after each e-commerce company provides experimental data, the platform system will conduct targeted adjustment training to obtain the coating color model of the corresponding e-commerce company; and for e-commerce companies that do not provide coating experimental data, the universal coating color model is directly used as the coating color model of the e-commerce company.

[0085] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0086] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An e-commerce platform system that uses virtual human technology to implement virtual trials of beauty products, characterized in that: include: Login module, information acquisition module, virtual module, interaction module, storage module; The login module confirms the identity information of the consumer and the e-commerce company and performs login management; The information acquisition module is used to acquire consumer information and beauty product information; The virtual module generates a corresponding virtual human model according to the consumer information; generates a corresponding virtual beauty product according to the beauty product information; and applies the virtual beauty product selected by the consumer to the virtual human model according to the consumer interaction information received by the interaction module, obtains a virtual trial model, and displays it through the interaction module; The storage module is used to store and update the identity information, as well as the consumer information corresponding to the identity information, the beauty product information, the virtual human model, the virtual beauty product, and the virtual trial model.

2. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 1, characterized in that: The login module includes: a consumer login unit and an e-commerce login unit, which are used for identity confirmation and login management of consumers and e-commerce respectively; The information acquisition module includes: a consumer information acquisition unit and a beauty product information acquisition unit, which respectively acquire the consumer information and the beauty product information; The virtual module includes a virtual human unit, a beauty product unit, and a virtual trial unit, which are respectively used to generate the virtual human model, the virtual beauty product, and the virtual trial model.

3. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 1, characterized in that: The consumer information includes consumer head picture data and facial key point data; The head picture data is a two-dimensional picture including the head of the consumer taken at multiple different angles; The facial key point data is obtained by moving the key points of the standard face model to determine the corresponding consumer face model after the consumer sets the length, width and height of the standard face model through the interaction module according to the consumer's own facial features, and then the information acquisition module reads the three-dimensional coordinates of the standard face model to obtain the facial key point data; The cosmetic products are divided into wearable cosmetic products and application cosmetic products; The beauty product information of the wearable beauty product includes: image data, three-dimensional scanning data, transparency, surface reflectivity, light scattering, and roughness; The beauty product information of the coating beauty product includes: color, color saturation, transparency, reflectivity, light scattering, and roughness.

4. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 3, characterized in that: Generating the corresponding virtual human model includes: Determine the shooting distance of each picture according to the shooting focal length of each picture; Extracting feature points from the head image data to determine facial key points contained in each image; Based on the shooting distance of each picture, the two-dimensional plane coordinates of the facial key points, and the three-dimensional coordinates of the facial key point data, the three-dimensional posture of the shooting device relative to the consumer face model when shooting each picture is solved to obtain a virtual camera posture set; Selecting two 3D postures that meet the set virtual camera spacing threshold and angle threshold from the virtual camera posture set as a group, performing binocular vision 3D reconstruction based on the corresponding two images, and obtaining a partial head 3D model of the consumer; Selecting multiple groups of three-dimensional postures, performing binocular vision three-dimensional reconstruction, and splicing the facial key point coordinates in the partial head three-dimensional model obtained by reconstruction to obtain a complete head three-dimensional model; The three-dimensional head model is subjected to surface smoothing processing to obtain the virtual human model.

5. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 4, characterized in that: Before the feature point extraction, the head image data is also preprocessed, including: firstly detecting the edge contour of the human body; then removing the background pixels according to the edge contour of the human body; and then converting the removed image from the RGB space to the HSV space.

6. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 3, characterized in that: The generating of the corresponding virtual beauty products comprises: For the wearable cosmetic product, determining a three-dimensional model of the wearable cosmetic product according to the three-dimensional scanning data, identifying a contour line of the three-dimensional model, and then determining a plurality of surfaces of the three-dimensional model according to the contour line; Performing contour recognition on the image data to determine the contour line of the wearable cosmetic product, and determining a plurality of surfaces according to the recognized contour line; Determining surface mapping data of the three-dimensional model according to a correspondence between the three-dimensional model and a surface determined by the image data; Determine the surface material data of the three-dimensional model according to the transparency, surface reflectivity, light scattering, and roughness of the wearable cosmetic product to obtain a wearable virtual cosmetic product; For the coated beauty products, surface color data is directly generated according to the color and color saturation of the coated beauty products, and surface material data is generated according to the transparency, reflectivity, light scattering, and roughness of the coated beauty products. The surface color data and surface material data are used as coated virtual beauty products.

7. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 6, characterized in that: The method of obtaining a virtual trial model comprises: For the wearable cosmetics product, the consumer directly splices the three-dimensional model of the wearable virtual cosmetics product onto the virtual human model through the interactive module to obtain the virtual trial model; For the coating type beauty products, the consumer first selects the coating area and coating amount on the virtual human model, then determines the color of the coated area after coating according to the coating color model, and uses the surface material data of the coating type virtual beauty products as the surface material of the coating area. Finally, the coating area selected on the virtual human model is modified according to the determined surface color and surface material to obtain the virtual trial model.

8. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 7, characterized in that: The coating color model is a three-layer fully connected neural network, the number of hidden layers in the three-layer neural network is 8, 8, and 1 respectively, the fully connected neural network uses the relu function as the activation function, and the loss function is the mean square error.

9. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 7, characterized in that: The coating color model is trained in the following manner: Obtain beauty products with different beauty product information, conduct coating experiments with different coating thicknesses on skin-like materials with different base colors, and collect the final coated area’s post-coating color; The color data, color saturation data, transparency data, reflectivity data, light scattering data, roughness data, coating thickness data, and coating base color data of the experimental cosmetics product are used as sample features, and the color data after coating is used as a sample label to form a sample data set, and the multiple sample data sets constitute a first sample data set; The first sample data set is used to train a fully connected neural network to obtain a universal coating color model.

10. The e-commerce platform system for realizing virtual trial of beauty products using virtual human technology according to claim 9, characterized in that: It also includes that the e-commerce company conducts the coating experiment on the beauty products it sells, collects the coated color of the final coated area, and organizes it to obtain a second sample data set, uses the second sample data set to adjust and train the universal coating color model to obtain the coating color model of each e-commerce product; and uses the Adam optimizer and 5-fold cross validation to perform the adjustment training.

Images