Garment sheet data generation method and device
The text encoder and generation network are used to convert clothing text data into pattern geometric features, which solves the problems of low manual design efficiency and high error rate in the traditional clothing industry, achieves fast and efficient clothing pattern design, and reduces pattern making costs.
Patent Information
- Application Number
- CN202310897588.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-07-20
- Publication Date
- 2025-01-21
AI Technical Summary
In the process of making plates, the traditional clothing industry has problems such as time-consuming and laborious, high error rate and low efficiency, which cannot meet the needs of rapid digitalization and highly customized personalization of clothing.
By obtaining the target clothing text data, using the text encoder and generation network to convert the text features into pattern geometric features, and outputting the pattern geometric data through the geometry decoder, automated pattern design is realized.
This method supports the rapid output of pattern geometric data. Compared with manual design, it is highly interactive, intelligent, fast and efficient in generation. It can greatly shorten the clothing modeling cycle and reduce the cost of pattern making.
Smart Images
Figure CN119337442A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of clothing pattern generation, and in particular to a method and device for generating clothing pattern data. Background Art
[0002] The traditional clothing industry has always faced the problem of pattern making. The traditional manual pattern design method lags behind the development of science and technology and cannot meet the needs of rapid clothing digitization and highly customized personalization. As the metaverse, digital fashion, and high-end private customization become the mainstream trends in clothing manufacturers, the demand for more specialization and diversification is also growing.
[0003] However, manual plate design is time-consuming and labor-intensive, and has problems such as high error rate and low efficiency, and can no longer meet production needs. Summary of the invention
[0004] In order to overcome the above-mentioned defects of the prior art, the present application provides a method and device for generating clothing pattern data.
[0005] The specific technical solutions are as follows:
[0006] A method for generating clothing pattern data, comprising:
[0007] Acquire target clothing text data and input the target clothing text data into a text encoder, and the text encoder outputs target clothing text features;
[0008] Inputting the target clothing text features into a generation network, and the generation network outputs the target pattern geometric features;
[0009] The target plate geometric features are input to a geometry decoder, and the geometry decoder outputs the target plate geometric data.
[0010] In one embodiment, the generation network is configured to obtain the target pattern geometric features based on the target garment text features and a text geometry dataset.
[0011] In one embodiment, the text geometry dataset is established in the following manner:
[0012] Get clothing text data and pattern geometry data;
[0013] Input the clothing text data into a text encoder to obtain clothing text features; input the pattern geometry data into a geometry encoder to obtain pattern geometry features;
[0014] The clothing text features with high correlation are linked to the pattern geometric features to obtain the text geometric data set.
[0015] In one embodiment, the target garment text features are input into a generation network, and the generation network outputs the target pattern geometric features, including:
[0016] Inputting the target clothing text features into a generation network;
[0017] The generation network outputs the target plate geometric features;
[0018] The generation network is iteratively generated until a preset number of iterations is reached to obtain the final geometric features of the target plate.
[0019] In one embodiment, the target plate geometric features are input into a geometry decoder, and the geometry decoder outputs the target plate geometric data, and the method also includes:
[0020] Obtaining target clothing line draft data and inputting the target clothing line draft data into an image encoder, or obtaining target clothing image data and inputting the target clothing image data into an image encoder; the image encoder outputs target clothing image features;
[0021] The target clothing image features are input into a generation network, and the generation network outputs the target pattern geometric features.
[0022] In one embodiment, acquiring target clothing image data and inputting the target clothing image data into an image encoder comprises:
[0023] Acquire target clothing image data;
[0024] Inputting the target clothing image data into a preprocessor, and the preprocessor outputs the target clothing line drawing data;
[0025] The target garment line drawing data is input into an image encoder.
[0026] In one embodiment, the generation network is configured to obtain the target pattern geometric features based on the target garment image features and the image geometry dataset;
[0027] The image geometry data set is established in the following manner: acquiring clothing image data and pattern geometry data; inputting the clothing image data into an image encoder to obtain clothing image features; inputting the pattern geometry data into a geometry encoder to obtain pattern geometry features; establishing a connection between the clothing image features and the pattern geometry features with a high degree of correlation to obtain the image geometry data set.
[0028] In one embodiment, the target garment image features are input into a generation network, and the generation network outputs the target pattern geometric features, including:
[0029] Inputting the target clothing image features into a generation network;
[0030] The generation network outputs the target plate geometric features;
[0031] The generation network is iteratively generated until a preset number of iterations is reached to obtain the final geometric features of the target plate.
[0032] A clothing pattern data generating device, comprising:
[0033] An acquisition module, used for acquiring target clothing text data and inputting the target clothing text data into a text encoder, and the text encoder outputs target clothing text features;
[0034] A generation module, used for inputting the target clothing text features into a generation network, and the generation network outputs the target pattern geometric features;
[0035] The output module is used to input the geometric features of the target plate into a geometric decoder, and the geometric decoder outputs the geometric data of the target plate.
[0036] In one embodiment, the generating module comprises:
[0037] An input module, used for inputting the target clothing text features into a generation network;
[0038] An updating module, used for the generating network to output the geometric features of the target plate;
[0039] The iteration module is used to perform iterative generation processing on the generation network until a preset number of iterations is reached to obtain the final geometric features of the target plate.
[0040] This application has at least the following beneficial effects:
[0041] The present application provides a method for generating clothing pattern data, including: obtaining target clothing text data and inputting the target clothing text data into a text encoder, the text encoder outputs target clothing text features; inputting the target clothing text features into a generation network, the generation network outputs target pattern geometric features; inputting the target pattern geometric features into a geometry decoder, the geometry decoder outputs target pattern geometric data. The clothing pattern data generation method provided by the present application supports text input and can quickly output corresponding pattern geometric data. Compared with manual design, the present application has high interactivity, strong intelligence, fast generation speed, high generation efficiency, and is easy to store and manage pattern geometric data, which can greatly shorten the clothing modeling cycle, thereby reducing the cost of clothing plate making.
[0042] The present application provides a device for generating clothing pattern data, including: an acquisition module for acquiring target clothing text data and inputting the target clothing text data into a text encoder, and the text encoder outputs target clothing text features; a generation module for inputting the target clothing text features into a generation network, and the generation network outputs target pattern geometric features; an output module for inputting the target pattern geometric features into a geometry decoder, and the geometry decoder outputs target pattern geometric data. The present application can quickly output corresponding pattern geometric data based on the user's target clothing text through the cooperation of the acquisition module, the generation module and the output module. Compared with manual design, the present application has high interactivity, strong intelligence, fast generation speed, high generation efficiency, and is easy to store and manage pattern geometric data, which can greatly shorten the clothing modeling cycle, thereby reducing the cost of clothing plate making. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0044] Figure 1 A first flow chart of the method for generating garment pattern data provided in Example 1;
[0045] Figure 2 A second flow chart of the method for generating garment pattern data provided in Example 1;
[0046] Figure 3 A third flow chart of the method for generating garment pattern data provided in Example 1;
[0047] Figure 4 A fourth flow chart of the method for generating garment pattern data provided in Example 1;
[0048] Figure 5 A fifth flow chart of the method for generating garment pattern data provided in Example 1;
[0049] Figure 6 A sixth flow chart of the method for generating garment pattern data provided in Example 1;
[0050] Figure 7 A schematic diagram of a first module of the garment pattern data generating device provided in Example 2;
[0051] Figure 8 This is a schematic diagram of the second module of the clothing pattern data generating device provided in Example 2.
[0052] Reference numerals:
[0053] 1-acquisition module; 2-generation module; 3-output module; 21-input module; 22-update module; 23-iteration module. DETAILED DESCRIPTION
[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0055] Example 1
[0056] like Figures 1 to 6 As shown, this embodiment provides a method for generating clothing pattern data, including:
[0057] Acquire target clothing text data and input the target clothing text data into a text encoder, and the text encoder outputs target clothing text features;
[0058] Input the target clothing text features into the generative network, and the generative network outputs the target pattern geometric features;
[0059] The target plate geometric features are input to the geometry decoder, and the geometry decoder outputs the target plate geometric data.
[0060] The clothing pattern data generation method provided in this embodiment supports text input and can quickly output corresponding pattern geometry data. Compared with manual design, this embodiment has high interactivity, strong intelligence, fast generation speed, high generation efficiency and easy storage and management of pattern geometry data, which can greatly shorten the clothing modeling cycle and thus reduce the cost of clothing pattern making.
[0061] Specifically, the target clothing text data refers to text describing clothing, such as a ripped denim jacket.
[0062] In one embodiment, the generation network is configured to obtain the target plate geometric features based on the target clothing text features and the text geometry data set. This embodiment establishes the text geometry data set in advance to achieve the pairing of clothing text features and plate geometry features, and then after obtaining the user's target clothing text features, the corresponding target plate geometry features can be quickly obtained according to the text geometry data set, with fast acquisition speed and high generation efficiency.
[0063] like Figure 4 As shown, in one embodiment, the text geometry dataset is established in the following manner:
[0064] Get clothing text data and pattern geometry data;
[0065] Input the clothing text data into the text encoder to obtain clothing text features; input the pattern geometry data into the geometry encoder to obtain pattern geometry features;
[0066] The clothing text features and pattern geometry features with high correlation are linked to obtain a text geometry dataset.
[0067] This embodiment uses a text encoder to encode clothing text data to obtain clothing text features, and uses a geometry encoder to encode plate geometry data to obtain plate geometry features, and associates clothing text features with plate geometry features to achieve consistent fusion representation of clothing text data and plate geometry data.
[0068] In one embodiment, the clothing text features with high correlation are linked to the pattern geometric features, including:
[0069] Calculate the correlation between each clothing text feature and each pattern geometric feature, and aggregate clothing text features and pattern geometric features with high correlation, and disperse clothing text features and pattern geometric features with low correlation;
[0070] The correlation between the clustered clothing text features and pattern geometric features is calculated again, and the clothing text features and pattern geometric features with high correlation are clustered again, and the clothing text features and pattern geometric features with low correlation are dispersed again;
[0071] Repeat the above recalculation steps until the preset number of iterations is reached;
[0072] The clothing text features and pattern geometric features that are finally gathered together are linked.
[0073] After multiple iterations of optimization, this embodiment finally establishes a connection between clothing text features and pattern geometric features with a high degree of correlation, so that the obtained pattern geometric features are consistent with the actual needs of the user.
[0074] like Figure 6 As shown, this embodiment pre-encodes the plate geometry data into low-dimensional plate geometry features through the text geometry data set, and finally decodes the obtained low-dimensional plate geometry features into target plate geometry data through the geometry decoder. Plate geometry features are more compact and effective than plate geometry data, avoid data redundancy, reduce the amount of calculation, improve the processing efficiency of the generation network, and shorten the time to generate target plate geometry data.
[0075] like Figure 2 , Figure 6 As shown, in one embodiment, the target clothing text features are input into the generation network, and the generation network outputs the target pattern geometric features, including:
[0076] Input the target clothing text features into the generation network;
[0077] Generate network outputs the target plate geometric features;
[0078] The generated network is iteratively generated until a preset number of iterations is reached to obtain the final target plate geometric features.
[0079] This embodiment iteratively generates the output target plate geometric features based on the target clothing text features and the generation network, and finally outputs the target plate geometric features that are more consistent with the user's actual needs, thereby improving the similarity between the subsequently generated target plate geometric data and the user's actual needs.
[0080] like Figure 3 , Figure 6 As shown, in one embodiment, the target plate geometric features are input to a geometric decoder, and the geometric decoder outputs the target plate geometric data, which also includes:
[0081] Obtaining target clothing line draft data and inputting the target clothing line draft data into an image encoder, or obtaining target clothing image data and inputting the target clothing image data into an image encoder; the image encoder outputs target clothing image features;
[0082] The target clothing image features are input into the generative network, and the generative network outputs the target pattern geometric features.
[0083] This embodiment supports multi-modal input. Users can obtain target plate geometry data not only by inputting text descriptions but also by inputting line drawings and clothing images, thereby achieving diversity in data input and a wide range of applications.
[0084] Specifically, the target clothing image data includes target clothing style image data.
[0085] In one embodiment, acquiring target clothing image data and inputting the target clothing image data into an image encoder includes:
[0086] Acquire target clothing image data;
[0087] Inputting the target clothing image data into a preprocessor, and the preprocessor outputs the target clothing line drawing data;
[0088] Input the target garment line drawing data into the image encoder.
[0089] This embodiment processes the target clothing image data through a preprocessor, converts the target clothing image data into target clothing line drawing data in advance, and inputs the obtained target clothing line drawing data into the image encoder to obtain the target clothing image features. Compared with the solution in which the image encoder directly processes the target clothing image data, this embodiment shortens the processing time of the image encoder and speeds up the time to obtain the target image features.
[0090] In one embodiment, the generation network is configured to obtain the target plate geometric features based on the target clothing image features and the image geometry data set. This embodiment establishes the image geometry data set in advance to achieve the pairing of clothing image features and plate geometry features, and then after obtaining the user's target clothing image features, the corresponding target plate geometry features can be quickly obtained according to the image geometry data set, with fast acquisition speed and high generation efficiency.
[0091] In one embodiment, the image geometry dataset is established by:
[0092] Obtain clothing image data and pattern geometry data;
[0093] Input the clothing image data into the image encoder to obtain clothing image features; input the pattern geometry data into the geometry encoder to obtain pattern geometry features;
[0094] The clothing image features with high correlation and the pattern geometric features are linked to obtain the image geometry dataset.
[0095] like Figure 5 As shown, this embodiment pre-encodes the plate geometry data into low-dimensional plate geometry features through the image geometry data set, and finally decodes the obtained low-dimensional plate geometry features into target plate geometry data through the geometry decoder. Plate geometry features are more compact and effective than plate geometry data, avoid data redundancy, reduce the amount of calculation, improve the processing efficiency of the generation network, and shorten the time to generate target plate geometry data.
[0096] like Figure 6 As shown, in one embodiment, the target garment image features are input into the generation network, and the generation network outputs the target plate geometric features, including:
[0097] Input the target clothing image features into the generation network;
[0098] Generate network outputs the target plate geometric features;
[0099] The generated network is iteratively generated until a preset number of iterations is reached to obtain the final target plate geometric features.
[0100] This embodiment performs iterative generation processing on the output target plate geometric features based on the target clothing image features and the generation network, and finally outputs the target plate geometric features that are more consistent with the user's actual needs, thereby improving the similarity between the subsequently generated target plate geometric data and the user's actual needs.
[0101] To sum up, this embodiment provides a method for generating clothing pattern data that meets multimodal input methods such as text description, line drawing and clothing image, has a high degree of intelligence, and reduces the usage threshold. This embodiment can quickly convert text description, line drawing, clothing image, etc. into pattern geometry data, realize a fast and accurate pattern design process, greatly improve the efficiency and quality of plate making, improve the efficiency of clothing digitization, reduce the difficulty of clothing digitization, and facilitate the further construction of digital clothing models.
[0102] Example 2
[0103] like Figure 7 As shown, this embodiment provides a clothing pattern data generating device, comprising:
[0104] An acquisition module 1 is used to acquire target clothing text data and input the target clothing text data into a text encoder, and the text encoder outputs target clothing text features;
[0105] A generation module 2 is used to input the target clothing text features into the generation network, and the generation network outputs the target pattern geometric features;
[0106] The output module 3 is used to input the geometric features of the target plate into the geometric decoder, and the geometric decoder outputs the geometric data of the target plate.
[0107] Through the cooperation of the acquisition module 21, the generation module 22 and the output module 23, this embodiment can quickly output the corresponding plate geometry data based on the user's target clothing text. Compared with manual design, the present application has high interactivity, strong intelligence, fast generation speed, high generation efficiency and easy storage and management of plate geometry data, which can greatly shorten the clothing modeling cycle and thus reduce the clothing plate making cost.
[0108] like Figure 8 As shown, in one embodiment, the generation module 22 includes:
[0109] An input module 21, used to input the target clothing text features into the generation network;
[0110] An updating module 22 is used to generate the geometric features of the target plate output by the network;
[0111] The iteration module 23 is used to perform iterative generation processing on the generated network until a preset number of iterations is reached to obtain the final target plate geometric features.
[0112] This embodiment, through the mutual cooperation of the input module 21, the update module 22 and the iteration module 23, iteratively generates the output target plate geometric features based on the target clothing text features and the generation network, and finally outputs the target plate geometric features that are more consistent with the user's actual needs, thereby improving the similarity between the subsequently generated target plate geometric data and the user's actual needs.
[0113] It should be understood by those skilled in the art that the above modules or steps of the present application can be implemented by a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, optionally, they can be implemented by a program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.
[0114] Note that the above are only preferred embodiments of the present application and the technical principles used. Those skilled in the art will understand that the present application is not limited to the specific embodiments herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present application. Therefore, although the present application is described in more detail through the above embodiments, the present application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the appended claims.
[0115] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.
Claims
1. A method for generating clothing pattern data, characterized in that: include: Acquire target clothing text data and input the target clothing text data into a text encoder, and the text encoder outputs target clothing text features; Inputting the target clothing text features into a generation network, and the generation network outputs the target pattern geometric features; The target plate geometric features are input to a geometry decoder, and the geometry decoder outputs the target plate geometric data.
2. The method for generating clothing pattern data according to claim 1, characterized in that: The generation network is configured to obtain the target pattern geometric features based on the target garment text features and a text geometry dataset.
3. The method for generating clothing pattern data according to claim 2, characterized in that: The text geometry dataset is established in the following way: Get clothing text data and pattern geometry data; Input the clothing text data into a text encoder to obtain clothing text features; input the pattern geometry data into a geometry encoder to obtain pattern geometry features; The clothing text features with high correlation are linked to the pattern geometric features to obtain the text geometric data set.
4. The method for generating clothing pattern data according to claim 1, characterized in that: The target garment text features are input into a generation network, and the generation network outputs the target pattern geometric features, including: Inputting the target clothing text features into a generation network; The generation network outputs the target plate geometric features; The generation network is iteratively generated until a preset number of iterations is reached to obtain the final geometric features of the target plate.
5. The method for generating clothing pattern data according to claim 1, characterized in that: The target plate geometric features are input into a geometric decoder, and the geometric decoder outputs the target plate geometric data, and the method also includes: Obtaining target clothing line draft data and inputting the target clothing line draft data into an image encoder, or obtaining target clothing image data and inputting the target clothing image data into an image encoder; the image encoder outputs target clothing image features; The target clothing image features are input into a generation network, and the generation network outputs the target pattern geometric features.
6. The method for generating clothing pattern data according to claim 5, characterized in that: Acquiring target clothing image data and inputting the target clothing image data into an image encoder comprises: Acquire target clothing image data; Inputting the target clothing image data into a preprocessor, and the preprocessor outputs the target clothing line drawing data; The target garment line drawing data is input into an image encoder.
7. The method for generating clothing pattern data according to claim 5, characterized in that: The generation network is configured to obtain the target pattern geometric features based on the target garment image features and the image geometry dataset; The image geometry data set is established in the following manner: acquiring clothing image data and pattern geometry data; inputting the clothing image data into an image encoder to obtain clothing image features; inputting the pattern geometry data into a geometry encoder to obtain pattern geometry features; establishing a connection between the clothing image features and the pattern geometry features with a high degree of correlation to obtain the image geometry data set.
8. The method for generating clothing pattern data according to claim 5, characterized in that: The target garment image features are input into a generation network, and the generation network outputs the target pattern geometric features, including: Inputting the target clothing image features into a generation network; The generation network outputs the target plate geometric features; The generation network is iteratively generated until a preset number of iterations is reached to obtain the final geometric features of the target plate.
9. A clothing pattern data generating device, characterized in that: include: An acquisition module, used for acquiring target clothing text data and inputting the target clothing text data into a text encoder, and the text encoder outputs target clothing text features; A generation module, used for inputting the target clothing text features into a generation network, and the generation network outputs the target pattern geometric features; The output module is used to input the geometric features of the target plate into a geometric decoder, and the geometric decoder outputs the geometric data of the target plate.
10. A clothing pattern data generating device according to claim 9, characterized in that: The generation module comprises: An input module, used for inputting the target clothing text features into a generation network; An updating module, used for the generating network to output the geometric features of the target plate; The iteration module is used to perform iterative generation processing on the generation network until a preset number of iterations is reached to obtain the final geometric features of the target plate.
Citation Information
Cited By
Text-driven garment template intelligent generation method
CN120197247A
A text-driven intelligent method for generating clothing patterns
CN120197247B