Intelligent clothing building method and system based on deep learning

Abstract

The invention relates to the technical field of intelligent clothing design, in particular to an intelligent clothing building method and system based on deep learning. Comprising a line draft scanning module, a partition module, a block integration module, a block adjustment module, an image generation module, a stereoscopic model import module, a wearing state generation module and a display terminal, the partition module is connected with the line draft scanning module, the block integration module is connected with the partition module, the block adjustment module is connected with the partition module, and the image generation module is connected with the display terminal. The image generation module is connected with the block integration module, the stereoscopic model import module is connected with the image generation module, the wearing state production module is connected with the image generation module, and the display terminal is connected with the image generation module; the design modification of the designated area can be completed according to the actual thought of a creator, meanwhile, rapid updating and presentation are carried out according to the corresponding modification content, the situation of repeated modification caused by large difference between a design scheme and actual display is avoided, and actual operation is more convenient.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent clothing design, and in particular to a method and system for intelligent clothing construction based on deep learning. Background Art

[0002] Image generation technology is a computer technology based on deep learning technology that automatically generates virtual images. In existing art creations such as clothing design, designers can use image generation technology to quickly generate a variety of artworks, thereby providing inspiration for creators.

[0003] Most of the existing intelligent clothing construction systems use adversarial learning methods to train models and generate images. The structure of the comparison process is multi-layer upsampling plus CNN feature extraction structure. Due to the use of cyclic adversarial training methods, the network can use the data set multiple times and fully learn the information in it, so that the quality of generated images is further improved.

[0004] In the existing intelligent clothing construction system, designers are generally required to read the design manuscripts after drawing them, and then generate corresponding image models based on the scanned line drawings to facilitate designers to observe more intuitively. However, with the diversification and detail of modern clothing design, the overall samples of most clothing design manuscripts are determined in advance, and then fine adjustments such as the matching size of some module designs, the size of patterns, and the pasting position are made according to the subsequent generated intuitive images. However, since the existing intelligent construction modules can only generate image models by scanning line drawings, when fine adjustments are required, designers are required to readjust the drawn line drawings for scanning in order to complete the secondary generation and modification, which is very inconvenient in actual operation. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for intelligent clothing construction based on deep learning, which can complete the design modification of the specified area according to the actual ideas of the creator, and quickly update and present according to the corresponding modification content, avoiding repeated modifications due to large differences between the design plan and the actual display, making it more convenient in actual operation.

[0006] To achieve the above-mentioned purpose, the present invention adopts a method and system for intelligent clothing construction based on deep learning, including a line draft scanning module, a partition module, a block integration module, a block adjustment module, an image generation module, a three-dimensional model import module, a wearing state generation module and a display terminal, wherein the partition module is connected to the line draft scanning module, the block integration module is connected to the partition module, the block adjustment module is connected to the partition module and to the block integration module, the image generation module is connected to the block integration module, the three-dimensional model import module is connected to the image generation module, the wearing state production module is connected to the image production module, and the display terminal is connected to the image generation module and to the wearing state generation module;

[0007] The line draft scanning module is used to scan the line draft after the creator's design is completed as a whole and generate a preliminary draft of the image model;

[0008] The partitioning module divides the generated draft clothing image into sections according to preset sections;

[0009] The block adjustment module edits, adjusts and integrates the block designs that need to be added and the corresponding panels that need to complete the corresponding block design modifications;

[0010] The block integration module is used to reconnect and integrate the partitioned panels after adjustment and modification;

[0011] The image generation module processes the processed overall clothing image to generate a complete design clothing image;

[0012] The three-dimensional model importing module produces a corresponding human body model according to the human body model parameters input by the creator;

[0013] The wearing state generating module converts the designed clothing image into a three-dimensional model, and cooperates with the imported human body model to complete a complete three-dimensional model of the clothing wearing state;

[0014] The display terminal displays the generated clothing image and the three-dimensional model wearing the clothing.

[0015] The line draft scanning module includes a line draft data extraction submodule and an image draft generation submodule, and the image draft generation submodule is connected to the line draft data extraction submodule;

[0016] The line draft data extraction submodule is used to scan the design line draft and extract the corresponding image modeling parameters;

[0017] The image draft generation submodule performs preliminary conversion on the designed clothing image according to the modeling parameters extracted by scanning.

[0018] Wherein, the partition module includes an image segmentation submodule and a segmentation plate arrangement submodule, the image segmentation submodule is connected to the image draft generation submodule, and the segmentation plate arrangement submodule is connected to the image segmentation submodule;

[0019] The image segmentation submodule performs modular segmentation on the draft of the produced clothing image according to the set standard;

[0020] The segmented plate arrangement submodule assigns values ​​and arranges the image plates after segmentation.

[0021] The block adjustment module includes a plate selection submodule, an adjustment scheme input submodule and a derivation submodule, wherein the plate selection submodule is connected to the segmented plate arrangement submodule, the adjustment scheme input submodule is connected to the plate selection submodule, and the derivation submodule is connected to the adjustment scheme input submodule;

[0022] The adjustment scheme input submodule selects the corresponding image blocks that need to be edited and modified according to the values ​​assigned to the segmented image blocks;

[0023] The adjustment scheme input submodule performs corresponding editing and modification on the selected image plate;

[0024] The export submodule outputs the modified corresponding image plate.

[0025] The block integration module includes a plate updating submodule, a plate synthesis submodule and an overall plate output submodule. The plate updating submodule is connected to the segmented plate arrangement submodule and to the export submodule. The plate synthesis submodule is connected to the plate updating submodule, and the overall plate output submodule is connected to the plate synthesis submodule.

[0026] The section updating submodule completes the replacement of the corresponding old section according to the modified new section;

[0027] The plate synthesis submodule performs splicing and integration of the overall image plates after the old plates are replaced;

[0028] The overall plate output submodule transmits the complete garment image after splicing and integration.

[0029] Wherein, the wearing state generation module includes a clothing image analysis submodule, a clothing three-dimensional modeling submodule and a fitting submodule, the clothing image analysis submodule is connected to the image generation module, the clothing three-dimensional modeling submodule is connected to the clothing image analysis submodule, and the fitting submodule is connected to the clothing three-dimensional modeling submodule;

[0030] The clothing image analysis submodule is used to analyze the generated complete clothing image and extract corresponding parameters;

[0031] The clothing three-dimensional modeling submodule completes the establishment of the clothing three-dimensional model according to the extracted image parameters;

[0032] The fitting submodule combines the generated human body model with the established clothing three-dimensional model to complete the establishment and output of the clothing wearing state three-dimensional model.

[0033] The three-dimensional model import module includes a model parameter input submodule and a model generation submodule, the model parameter input submodule is connected to the model generation submodule, and the model generation submodule is connected to the fitting submodule;

[0034] The model parameter input submodule is used for designers to input the parameters of the established human body model according to their own needs;

[0035] The model generation submodule completes the establishment and output of the corresponding human body model according to the three-dimensional parameters input by the designer.

[0036] The clothing intelligent construction method and system based on deep learning of the present invention firstly converts the designer's line draft through the line draft data extraction submodule provided in the line draft scanning module in cooperation with the image draft generation submodule. When collecting data and converting images of the line draft, multiple layers of conv-block can be stacked to enhance the feature expression ability of the feature layer and reduce the problems of gradient disappearance and gradient diffusion during training. The lsGAN mechanism is introduced during training to penalize the gradient and enhance the stability of model training. After the image is initially generated, the generated clothing picture can be cut and divided into blocks through the partitioning module provided. Specifically, The block category can be selected according to the actual operation of the user. For example, the block can be divided according to the clothing design elements, or according to the clothing design blocks. The multiple blocks that have been sorted and completed can facilitate the user to directly operate the selected blocks through the block adjustment module when the user needs to modify and edit the content of a certain module, such as adding new elements or changing the position of elements and changing the size of the selected module and other parameters that need to be fine-tuned. After editing and modifying a specified single block or multiple blocks, the block integration module will replace the modified content with the corresponding block of the first draft image produced first, and then replace the replaced The block is spliced ​​and integrated with the other modules that retain the initial content, and finally a complete clothing image is generated through the image generation module. The generated complete clothing image can be directly viewed by the display terminal for designers. At the same time, the generated complete clothing image can be further processed by the three-dimensional model import module to generate a three-dimensional model of the corresponding designed clothing. Then, the generated clothing three-dimensional model is combined with a human body model of corresponding size in cooperation with the three-dimensional model import module and the wearing state generation module, so as to simulate the effect of displaying the actual body of the clothing. It is worth noting that the user can make real-time multiple adjustments and modifications to the selected module according to the block adjustment module during actual operation. When making multiple modifications, as the modification data is continuously adjusted, the block integration module will also be adjusted accordingly on the basis of the original manuscript. Finally, the clothing image and clothing wearing effect after each adjustment can be displayed through the display terminal, which is convenient for users to make continuous and minor adjustments and modifications, and there is no need to re-modify the whole and then scan and process, so that the design modification of the specified area can be completed according to the actual idea of ​​the creator, and the corresponding modification content is quickly updated and presented, so as to avoid repeated modifications due to large differences between the design scheme and the actual display, making it more convenient in actual operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 It is a structural schematic diagram of the clothing intelligent building system based on deep learning of the present invention.

[0039] Figure 2 It is a structural schematic diagram of the partitioning module, the block integration module and the block adjustment module of the present invention.

[0040] Figure 3 It is a structural schematic diagram of the wearing state generating module and the three-dimensional model importing module of the present invention.

[0041] Figure 4 It is a flow chart of the clothing intelligent construction method based on deep learning of the present invention.

[0042] In the figure: 1-line draft scanning module, 2-partition module, 3-block integration module, 4-block adjustment module, 5-image generation module, 6-3D model import module, 7-wearing state generation module, 8-display terminal, 11-line draft data extraction submodule, 12-image draft generation submodule, 21-image segmentation submodule, 22-segmentation plate arrangement submodule, 31-plate update submodule, 32-plate synthesis submodule, 33-overall plate output submodule, 41-plate selection submodule, 42-adjustment plan input submodule, 43-export submodule, 61-model parameter input submodule, 62-model generation submodule, 71-clothing image analysis submodule, 72-clothing 3D modeling submodule, 73-fitting submodule. DETAILED DESCRIPTION

[0043] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0044] In the description of the present invention, it should be understood that “plurality” means two or more than two, unless otherwise clearly and specifically defined.

[0045] See also Figures 1 to 3The present invention provides a method and system for intelligent clothing construction based on deep learning, including a line draft scanning module 1, a partition module 2, a block integration module 3, a block adjustment module 4, an image generation module 5, a three-dimensional model import module 6, a wearing state generation module 7 and a display terminal 8, wherein the partition module 2 is connected to the line draft scanning module 1, the block integration module 3 is connected to the partition module 2, the block adjustment module 4 is connected to the partition module 2 and is connected to the block integration module 3, the image generation module 5 is connected to the block integration module 3, the three-dimensional model import module 6 is connected to the image generation module 5, the wearing state production module is connected to the image production module, and the display terminal 8 is connected to the image generation module 5 and is connected to the wearing state generation module 7;

[0046] The line draft scanning module 1 is used to scan the line draft after the creator's design is completed as a whole and generate a preliminary draft of the image model;

[0047] The partitioning module 2 is configured to partition the generated draft clothing image according to preset sections;

[0048] The block adjustment module 4 edits, adjusts and integrates the block designs that need to be added and the corresponding panels that need to complete the corresponding block design modification;

[0049] The block integration module 3 is used to reconnect and integrate the partitioned plates after adjustment and modification;

[0050] The image generation module 5 processes the processed overall clothing image to generate a complete design clothing image;

[0051] The three-dimensional model importing module 6 produces a corresponding human body model according to the human body model parameters input by the creator;

[0052] The wearing state generating module 7 converts the designed clothing image into a three-dimensional model, and cooperates with the imported human body model to complete a complete three-dimensional model of the clothing wearing state;

[0053] The display terminal 8 displays the generated clothing image and the three-dimensional model wearing the clothing.

[0054] For further information, see Figure 2 The line draft scanning module 1 includes a line draft data extraction submodule 11 and an image draft generation submodule 12, and the image draft generation submodule 12 is connected to the line draft data extraction submodule 11;

[0055] The line drawing data extraction submodule 11 is used to scan the design line drawing and extract the corresponding image modeling parameters;

[0056] The image draft generation submodule 12 performs preliminary conversion on the designed clothing image according to the modeling parameters extracted by scanning.

[0057] In this embodiment, the line draft data extraction submodule 11 provided in the line draft scanning module 1 cooperates with the image draft generation submodule 12 to perform a preliminary conversion on the designer's line draft. When performing data collection and image conversion on the line draft, multiple layers of conv-block can be stacked to enhance the feature expression ability of the feature layer and reduce the gradient disappearance and gradient diffusion problems during training. The lsGAN mechanism is introduced during training to penalize the gradient and enhance the stability of model training. After the image is initially generated, the generated clothing image can be cut and divided into blocks by the partitioning module 2. The specific cutting and block categories can be based on The user can select according to the actual operation situation, for example, the section can be divided according to the clothing design elements, or according to the clothing design section. The multiple sections after these divisions and sorting can facilitate the user to directly operate the selected section through the block adjustment module 4 when the user needs to modify and edit the content of a certain module, such as adding new elements or changing the position of elements and changing the size of the selected module and other parameters that need to be fine-tuned. After editing and modifying the specified single section or multiple sections, the block integration module 3 will replace the modified content with the corresponding section of the first draft image produced first, and then replace the replaced section with the remaining maintained sections. The modules of the initial content are spliced ​​and integrated, and finally a complete clothing image is generated through the image generation module 5. The generated complete clothing image can be directly viewed by the display terminal 8 for designers. At the same time, the generated complete clothing image can be further processed by the three-dimensional model importing module 6 to generate a three-dimensional model of the corresponding designed clothing. Then, the three-dimensional model importing module 6 and the wearing state generating module 7 are used to combine the generated clothing three-dimensional model with a human body model of a corresponding size, so as to simulate and display the effect of actually wearing the clothing. It is worth noting that the user can adjust the selected module according to the block adjustment module 4 during actual operation. Multiple adjustments and modifications can be made in real time. When making multiple modifications, as the modification data is continuously adjusted, the block integration module 3 will also make corresponding continuous adjustments based on the original manuscript. Finally, the clothing image and clothing wearing effect after each adjustment can be displayed through the display terminal 8, which is convenient for users to make continuous and small adjustments and modifications. There is no need to re-modify the entire area and then scan and process it. This enables the design modification of the specified area to be completed according to the actual ideas of the creator, and at the same time, it is quickly updated and presented according to the corresponding modification content, avoiding repeated modifications due to large differences between the design plan and the actual display, making it more convenient in actual operation.

[0058] For further information, see Figure 2 The partition module 2 includes an image segmentation submodule 21 and a segmentation plate arrangement submodule 22, the image segmentation submodule 21 is connected to the image draft generation submodule 12, and the segmentation plate arrangement submodule 22 is connected to the image segmentation submodule 21;

[0059] The image segmentation submodule 21 performs modular segmentation on the draft of the produced clothing image according to the set standard;

[0060] The segmented block arrangement submodule 22 assigns values ​​and arranges the image blocks after segmentation.

[0061] For further information, see Figure 2 The block adjustment module 4 includes a plate selection submodule 41, an adjustment scheme input submodule 42 and a derivation submodule 43. The plate selection submodule 41 is connected to the segmented plate arrangement submodule 22, the adjustment scheme input submodule 42 is connected to the plate selection submodule 41, and the derivation submodule 43 is connected to the adjustment scheme input submodule 42;

[0062] The adjustment scheme input submodule 42 selects the corresponding image blocks that need to be edited and modified according to the values ​​assigned to the segmented image blocks;

[0063] The adjustment scheme input submodule 42 performs corresponding editing and modification on the selected image plate;

[0064] The export submodule 43 outputs the modified corresponding image block.

[0065] For further information, see Figure 2 The block integration module includes a plate updating submodule 31, a plate synthesis submodule 32 and an overall plate output submodule 33. The plate updating submodule 31 is connected to the segmented plate arrangement submodule and is connected to the export submodule 43. The plate synthesis submodule 32 is connected to the plate updating submodule 31, and the overall plate output submodule 33 is connected to the plate synthesis submodule 32.

[0066] The section updating submodule 31 completes the replacement of the corresponding old section according to the modified new section;

[0067] The plate synthesis submodule 32 performs splicing and integration of the whole image plate after the old plate is replaced;

[0068] The overall plate output submodule 33 transmits the complete garment image after splicing and integration.

[0069] In this embodiment, the generated draft clothing image is cut according to the method set by the user through the image segmentation submodule 21. The specific cutting and partitioning method can be set in advance according to the actual design requirements of the user. After the image is cut according to the set rules, the multiple image panels formed by the segmentation will be assigned and sorted by the segmentation panel arrangement submodule 22 to facilitate the user to quickly select the panels that need to be edited. Therefore, the user only needs to complete the selection of the corresponding editing panel according to the assignment of the comparison image panel through the panel selection submodule 41, and then modify and edit the selected panel to be edited through the adjustment scheme input submodule 42, and finally output the corresponding panel after the modification through the export submodule 43. As for the panel update submodule 31, it is responsible for replacing the modified panel with the corresponding draft clothing image according to the number assignment, and then cooperates with the panel synthesis submodule 32 to generate a new draft of the clothing image after the modification, and finally outputs the new draft of the clothing image after the modification through the overall panel output submodule 33.

[0070] For further information, see Figure 3 The wearing state generating module 7 includes a clothing image analyzing submodule 71, a clothing three-dimensional modeling submodule 72 and a fitting submodule 73, wherein the clothing image analyzing submodule 71 is connected to the image generating module 5, the clothing three-dimensional modeling submodule 72 is connected to the clothing image analyzing submodule 71, and the fitting submodule 73 is connected to the clothing three-dimensional modeling submodule 72;

[0071] The clothing image analysis submodule 71 is used to analyze the generated complete clothing image and extract corresponding parameters;

[0072] The clothing three-dimensional modeling submodule 72 completes the establishment of a three-dimensional model of the clothing according to the extracted image parameters;

[0073] The fitting submodule 73 combines the generated human body model with the established clothing three-dimensional model to complete the establishment and output of the clothing wearing state three-dimensional model.

[0074] For further information, see Figure 3 The three-dimensional model importing module 6 includes a model parameter input submodule 61 and a model generation submodule 62, wherein the model parameter input submodule 61 is connected to the model generation submodule 62, and the model generation submodule 62 is connected to the fitting submodule 73;

[0075] The model parameter input submodule 61 is used for designers to input the parameters of the established human body model according to their own needs;

[0076] The model generation submodule 62 completes the establishment and output of the corresponding human body model according to the three-dimensional parameters input by the designer.

[0077] In this embodiment, the model parameter input submodule 61 and the model generation submodule 62 are modules for generating a human body model. The parameters such as human height input by the model parameter input submodule 61 cooperate with the model generation submodule 62 to complete the establishment of a corresponding type of human body three-dimensional model, and then the fitting submodule 73 is used to cooperate with the established clothing three-dimensional model to complete the display of the effect of clothing on the body. The clothing three-dimensional model can be established by the clothing image analysis submodule 71 and the clothing three-dimensional modeling submodule 72. The clothing image generated by the image generation module 5 is specified by the clothing image analysis submodule 71. The parameters are analyzed and extracted, and then the three-dimensional model is established through the clothing three-dimensional modeling submodule 72 according to the extracted parameters. The specific conversion into a three-dimensional model is through the extended application of a generative adversarial network (GAN) and a variational autoencoder (VAE). These networks were originally used for tasks such as image generation, and are now extended to generate a three-dimensional model from a single plane image. By training on a large number of three-dimensional models and corresponding two-dimensional image data, the model learns the mapping relationship from two-dimensional image features to three-dimensional model structures. First, a large amount of three-dimensional model data and corresponding two-dimensional rendered images are prepared to construct a training data set, and then a network architecture based on GAN or VAE is designed and trained on the training data set. During the training process, the network continuously adjusts parameters to improve the accuracy of generating a three-dimensional model from a two-dimensional image. Finally, the plane image to be modeled is input into the trained network, and the corresponding three-dimensional model is output. The independent human body model and clothing three-dimensional model can be combined through the fitting submodule 73. The specific method requires that the clothing be regarded as a cloth with physical properties (such as elasticity, weight, friction, etc.), and the hanging, stretching, folding, etc. of the cloth on the three-dimensional model are simulated by the physical engine. Then, in the interaction process between the clothing and the three-dimensional model, it is necessary to detect the collision between the clothing model and the three-dimensional model in real time, and make corresponding adjustments according to the collision situation, which includes judging the location, area and strength of the collision, and avoiding unreasonable penetration by changing the shape or position of the clothing model. For example, in the existing virtual fitting system, when the user chooses to try on a pair of tights, the system will detect the collision between the pants model and the legs of the human body model in real time. If it is found that the pants model penetrates the human body model in some parts (such as the knees), the system will automatically adjust the local shape or position of the pants model so that the pants can fit naturally on the legs without penetration.

[0078] See also Figure 4A clothing intelligent construction method based on deep learning, using the clothing intelligent construction system based on deep learning, comprises the following steps:

[0079] S1: firstly, the line draft is scanned by the line draft scanning module 1, and a clothing image draft file is generated, and then the generated clothing image draft file is separated by a preset template by the partitioning module 2;

[0080] S2: After the clothing image draft file module is separated, the corresponding block is selected and adjusted and modified by the block adjustment module 4, and then the selected and modified block is replaced and updated by the block integration module 3, and finally integrated into a modified new clothing image file;

[0081] S3: The generated clothing image file is processed and exported through the image generation module. The user can directly view the generated clothing image file through the display terminal 8. At the same time, the clothing image file can also be three-dimensionally modeled through the subsequent three-dimensional model import module 6 and the wearing state generation module 7, and then the clothing wearing state image can be displayed in combination with the generated human body model.

[0082] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A clothing intelligent construction system based on deep learning, characterized in that: It includes a line draft scanning module, a partitioning module, a block integration module, a block adjustment module, an image generation module, a three-dimensional model importing module, a wearing state generation module and a display terminal, wherein the partitioning module is connected to the line draft scanning module, the block integration module is connected to the partitioning module, the block adjustment module is connected to the partitioning module and is connected to the block integration module, the image generation module is connected to the block integration module, the three-dimensional model importing module is connected to the image generation module, the wearing state production module is connected to the image production module, and the display terminal is connected to the image generation module and is connected to the wearing state generation module; The line draft scanning module is used to scan the line draft after the creator's design is completed as a whole and generate a preliminary draft of the image model; The partitioning module divides the generated draft clothing image into sections according to preset sections; The block adjustment module edits, adjusts and integrates the block designs that need to be added and the corresponding panels that need to complete the corresponding block design modifications; The block integration module is used to reconnect and integrate the partitioned panels after adjustment and modification; The image generation module processes the processed overall clothing image to generate a complete design clothing image; The three-dimensional model importing module produces a corresponding human body model according to the human body model parameters input by the creator; The wearing state generating module converts the designed clothing image into a three-dimensional model, and cooperates with the imported human body model to complete a complete three-dimensional model of the clothing wearing state; The display terminal displays the generated clothing image and the three-dimensional model wearing the clothing.

2. The clothing intelligent construction system based on deep learning according to claim 1, characterized in that: The line draft scanning module includes a line draft data extraction submodule and an image draft generation submodule, and the image draft generation submodule is connected to the line draft data extraction submodule; The line draft data extraction submodule is used to scan the design line draft and extract the corresponding image modeling parameters; The image draft generation submodule performs preliminary conversion on the designed clothing image according to the modeling parameters extracted by scanning.

3. The clothing intelligent construction system based on deep learning as claimed in claim 2, characterized in that: The partitioning module includes an image segmentation submodule and a segmentation plate arrangement submodule, the image segmentation submodule is connected to the image draft generation submodule, and the segmentation plate arrangement submodule is connected to the image segmentation submodule; The image segmentation submodule performs modular segmentation on the draft of the produced clothing image according to the set standard; The segmented plate arrangement submodule assigns values ​​and arranges the image plates after segmentation.

4. The clothing intelligent construction system based on deep learning as claimed in claim 3, characterized in that: The block adjustment module includes a plate selection submodule, an adjustment scheme input submodule and a derivation submodule, wherein the plate selection submodule is connected to the segmented plate arrangement submodule, the adjustment scheme input submodule is connected to the plate selection submodule, and the derivation submodule is connected to the adjustment scheme input submodule; The adjustment scheme input submodule selects the corresponding image blocks that need to be edited and modified according to the values ​​assigned to the segmented image blocks; The adjustment scheme input submodule performs corresponding editing and modification on the selected image plate; The export submodule outputs the modified corresponding image plate.

5. The clothing intelligent construction system based on deep learning as claimed in claim 4, characterized in that: The block integration module includes a plate updating submodule, a plate synthesis submodule and an overall plate output submodule, wherein the plate updating submodule is connected to the segmented plate arrangement submodule and to the export submodule, the plate synthesis submodule is connected to the plate updating submodule, and the overall plate output submodule is connected to the plate synthesis submodule; The section updating submodule completes the replacement of the corresponding old section according to the modified new section; The plate synthesis submodule performs splicing and integration of the overall image plates after the old plates are replaced; The overall plate output submodule transmits the complete garment image after splicing and integration.

6. The clothing intelligent construction system based on deep learning as claimed in claim 5, characterized in that: The wearing state generation module includes a clothing image analysis submodule, a clothing three-dimensional modeling submodule and a fitting submodule, wherein the clothing image analysis submodule is connected to the image generation module, the clothing three-dimensional modeling submodule is connected to the clothing image analysis submodule, and the fitting submodule is connected to the clothing three-dimensional modeling submodule; The clothing image analysis submodule is used to analyze the generated complete clothing image and extract corresponding parameters; The clothing three-dimensional modeling submodule completes the establishment of the clothing three-dimensional model according to the extracted image parameters; The fitting submodule combines the generated human body model with the established clothing three-dimensional model to complete the establishment and output of the clothing wearing state three-dimensional model.

7. The clothing intelligent construction system based on deep learning according to claim 6, characterized in that: The three-dimensional model importing module includes a model parameter input submodule and a model generation submodule, wherein the model parameter input submodule is connected to the model generation submodule, and the model generation submodule is connected to the fitting submodule; The model parameter input submodule is used for designers to input the parameters of the established human body model according to their own needs; The model generation submodule completes the establishment and output of the corresponding human body model according to the three-dimensional parameters input by the designer.

8. A method for intelligent clothing construction based on deep learning, using the intelligent clothing construction system based on deep learning as claimed in claim 1, characterized in that: The following steps are included: Firstly, the line draft is scanned by the line draft scanning module to generate a clothing image draft file, and then the generated clothing image draft file is separated by a preset template by the partitioning module; After the module separation of the draft clothing image file is completed, the corresponding blocks are selected and adjusted and modified by the block adjustment module, and then the selected and modified blocks are replaced and updated by the block integration module, and finally integrated into a modified new clothing image file; The generated clothing image file is processed and exported through the image generation module, and the user can directly view the generated clothing image file through the display terminal. At the same time, the clothing image file can also be three-dimensionally modeled through the subsequent three-dimensional model import module in conjunction with the wearing state generation module, and then the clothing wearing state image can be displayed in conjunction with the generated human body model.

Images