Three-dimensional costume design drawing generation method and device and electronic equipment

Abstract

The invention provides a three-dimensional costume design drawing generation method and device and electronic equipment, and belongs to the technical field of costume design. In the method, first text information is used as input of a three-dimensional garment design drawing generation model, and the first text information describes color information, style information, material information and detail information of a first three-dimensional garment design drawing expected by a user by adopting a natural language; and performing prediction processing on the first text information through the three-dimensional costume design drawing generation model to obtain the first three-dimensional costume design drawing, thereby realizing efficient conversion from a natural language to the three-dimensional costume design drawing. The three-dimensional costume design drawing generation model takes text information expressed by a natural language as input, so that the convenience and the flexibility of expressing a design intention by a user are remarkably improved. Meanwhile, the color, style, material and details of the garment are accurately expressed by using the first text information, so that an information gap between the design intention and the generation result is effectively reduced, and the authenticity of the three-dimensional garment design drawing is improved.

Description

Technical Field

[0001] The present application relates to the technical field of clothing design, and in particular to a method, device and electronic equipment for generating three-dimensional clothing design drawings. Background Art

[0002] With the rapid development of digital design technology, the automatic generation of 3D clothing design drawings has become an important direction for the intelligentization of the clothing industry. Through computer-aided design, designers can more efficiently realize their creative expression, shorten the design cycle, and meet personalized needs.

[0003] Related technologies typically employ image prediction techniques based on convolutional neural networks (CNNs) combined with principal component analysis (PCA) to simplify clothing panel data, or employ differentiable simulation optimization techniques to restore clothing material properties and pattern details. These methods improve the automation and refinement of design to a certain extent, but they often rely on complex parameter settings and image processing processes, lack intuitive and diverse user interaction methods, and struggle to meet the convenience and flexibility users require when expressing their design intent. This results in a poor user experience and limited applicability. Furthermore, existing technologies have yet to achieve direct and efficient interaction with natural language, leading to an information gap between design intent and generated results.

[0004] Therefore, in the technology of automatic generation of three-dimensional clothing design drawings, the complexity of user interaction and the inconvenience of operation, as well as the poor interaction effect between natural language and design models, have become problems that need to be solved urgently. Summary of the Invention

[0005] The present application provides a method, device and electronic device for generating three-dimensional clothing design drawings to solve the problems of complex user interaction and inconvenient operation in existing three-dimensional clothing design drawing automatic generation technology, as well as poor interaction between natural language and design models.

[0006] In a first aspect, the present application provides a method for generating a three-dimensional clothing design drawing, the method comprising:

[0007] Obtaining first text information, the first text information being used to describe color information, style information, material information, and detail information of a desired first three-dimensional clothing design drawing; wherein the detail information is used to indicate at least one of a pattern, a collar type, a sleeve type, a skirt hem, and a trouser leg related to the first three-dimensional clothing design drawing;

[0008] The first text information is predicted and processed based on a three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing; wherein the three-dimensional clothing design drawing generation model is obtained by training three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples, and the second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design drawing samples.

[0009] In one possible design, the predictive processing of the first text information based on the three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing includes:

[0010] Processing the first text information to obtain a first text guidance vector;

[0011] Based on the three-dimensional clothing design drawing generation model, cross-attention processing is performed on the starting word and the first text guide vector to obtain a target word sequence;

[0012] Based on the target word sequence, the first three-dimensional clothing design drawing is obtained and output.

[0013] In one possible design, performing cross-attention processing on the starting word and the first text guide vector to obtain a target word sequence includes:

[0014] performing a cross-attention process on the starting word-unit and the first text guide vector to obtain a first word-unit;

[0015] concatenating the starting word and the first word to obtain a first word sequence;

[0016] Setting the first word-gram sequence as the starting word-gram, and repeatedly performing the above cross-attention processing steps until an ending word-gram is obtained;

[0017] The end word and the start word corresponding to the end word are concatenated to obtain the target word sequence.

[0018] In one possible design, performing cross-attention processing on the starting word-unit and the first text guide vector to obtain the first word-unit includes:

[0019] The starting word is processed based on a position encoder, a parameter encoder, and a value encoder respectively to obtain a position latent vector, a parameter latent vector, and a value latent vector corresponding to the starting word; wherein the position encoder, the parameter encoder, and the value encoder are obtained by training the three-dimensional clothing design drawing sample; the position latent vector corresponding to the starting word is used to indicate the position of the starting word on the first three-dimensional clothing design drawing; the parameter latent vector corresponding to the starting word is used to indicate the parameter type of the starting word; and the value latent vector corresponding to the starting word is used to indicate the numerical information corresponding to the starting word;

[0020] Concatenating the starting word, and the position latent vector, the parameter latent vector, and the value latent vector corresponding to the starting word to obtain a second word sequence;

[0021] Cross-attention processing is performed on the second word-gram sequence and the first text guide vector to obtain the first word-gram.

[0022] In one possible design, before obtaining the first text information, the method further includes:

[0023] Acquire a training sample set, wherein the training sample set includes a plurality of training samples, and the training samples are the three-dimensional clothing design drawing samples and the second text information corresponding to the three-dimensional clothing design drawing samples;

[0024] Based on the multiple training samples, the initial three-dimensional clothing design drawing generation model is trained to obtain the three-dimensional clothing design drawing generation model.

[0025] In one possible design, the initial three-dimensional clothing design drawing generation model is trained based on the multiple training samples to obtain the three-dimensional clothing design drawing generation model, including:

[0026] For any training sample among the plurality of training samples, obtaining a second text guide vector based on the second text information corresponding to the three-dimensional clothing design sample;

[0027] Performing cross-attention processing on the starting word and the second text guide vector to obtain a third word sequence corresponding to the three-dimensional clothing design sample;

[0028] Determining a fourth word sequence corresponding to the three-dimensional clothing design sample using a preset rule;

[0029] A loss function is constructed based on the difference between the third word sequence and the fourth word sequence, and the initial three-dimensional clothing design drawing generation model is updated based on the loss function to obtain the three-dimensional clothing design drawing generation model.

[0030] In a possible design, the determining of the fourth word sequence corresponding to the three-dimensional clothing design sample by using a preset rule includes:

[0031] Determine the rotation quaternion and translation vector corresponding to each of the multiple polygonal panels corresponding to the three-dimensional clothing design sample, as well as the edge information and stitching information corresponding to the multiple edges of each polygonal panel; wherein the rotation quaternion is used to indicate the rotation angle of the polygonal panel relative to the global coordinate system; the translation vector is used to indicate the position offset of the polygonal panel relative to the global coordinate system; the edge information is used to indicate the starting point, end point and Bezier curve control point of the edge; the stitching information is used to indicate whether to stitch the edge and the position of the target edge to be stitched with the edge;

[0032] The rotation quaternions and translation vectors corresponding to the multiple polygonal panels, and the edge information and stitching information corresponding to the multiple edges of each polygonal panel are spliced ​​to obtain the fourth word sequence.

[0033] In a second aspect, the present application provides a three-dimensional clothing design drawing generation device, comprising:

[0034] an acquisition module, configured to acquire first text information, wherein the first text information is used to describe color information, style information, material information, and detail information of a desired first three-dimensional clothing design drawing; wherein the detail information is used to indicate at least one of a pattern, a collar shape, a sleeve shape, a skirt hem, and a trouser leg related to the first three-dimensional clothing design drawing;

[0035] An output module is used to predict and process the first text information based on a three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing; wherein the three-dimensional clothing design drawing generation model is trained by three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples, and the second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design drawing samples.

[0036] In a third aspect, the present application provides an electronic device, comprising: a memory and at least one processor;

[0037] The memory stores computer-executable instructions;

[0038] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the method described in the first aspect or various possible designs of the first aspect.

[0039] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer execution instructions are stored. When the computer execution instructions are executed, the method described in the first aspect or various possible designs of the first aspect is implemented.

[0040] In a fifth aspect, the present application provides a computer program product, which includes a computer program code. When the computer program code runs on a computer, the computer implements the method described in the first aspect or various possible designs of the first aspect.

[0041] In the sixth aspect, the present application provides a chip comprising: an interface circuit and a logic circuit, wherein the interface circuit is used to receive signals from other chips outside the chip and transmit them to the logic circuit, or to send signals from the logic circuit to other chips outside the chip, and the logic circuit is used to implement the method described in the first aspect or various possible designs of the first aspect.

[0042] The embodiment of the present application provides a three-dimensional clothing design drawing generation method, device and electronic device, in which the first text information is used as the input of the three-dimensional clothing design drawing generation model, the first text information uses natural language to describe the color information, style information, material information and detail information of the first three-dimensional clothing design drawing that the user expects to obtain, and the first text information is predicted and processed by the three-dimensional clothing design drawing generation model to obtain the corresponding first three-dimensional clothing design drawing, thereby realizing efficient conversion from natural language to three-dimensional clothing design drawings. The three-dimensional clothing design drawing generation model uses text information expressed in natural language as input, which significantly improves the convenience and flexibility of users in expressing their design intentions. At the same time, the first text information is used to accurately express the color, style, material and details of the clothing, effectively narrowing the information gap between the design intention and the generated results, and improving the authenticity and professionalism of the three-dimensional clothing design drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a flow chart of a method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application;

[0044] Figure 2 A flowchart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application;

[0045] Figure 3 A flowchart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application;

[0046] Figure 4 A schematic flow chart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application;

[0047] Figure 5 A schematic structural diagram of a three-dimensional clothing design drawing generating device provided in an embodiment of the present application;

[0048] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification, claims and drawings of this application are intended to cover non-exclusive inclusions.

[0051] References to "embodiments" herein mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiment" in various places in the specification does not necessarily refer to the same embodiment, nor does it necessarily refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0052] The term "and / or" in this document simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists, A and B can exist at the same time, and B exists. Additionally, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0053] In addition, the terms "first", "second", etc. in the description and claims of this application or the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order, and may explicitly or implicitly include one or more such features.

[0054] In the description of this application, unless otherwise specified, "multiple" and "at least two" mean more than two (including two). Similarly, "multiple groups" and "at least two groups" mean more than two groups (including two).

[0055] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "connected" and "connect" should be understood in a broad sense. For example, "connected" or "connected" can refer not only to physical connections, but also to electrical connections or signal connections. For example, it can be a direct connection, i.e., a physical connection, or an indirect connection through at least one intermediate element, as long as circuit connectivity is achieved. It can also refer to internal connectivity between two elements. Signal connection can refer not only to signal connection through circuits, but also to signal connection through media, such as radio waves. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0056] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings. It should be noted that different technical features in the present application can be combined with each other in the absence of conflict.

[0057] First, the terms involved in one or more embodiments of this specification are explained.

[0058] A token is the smallest unit or basic element used to represent the processed text in the natural language processing process. A token can be a single character or a sequence of multiple characters.

[0059] Below, the implementation background of the technical solution provided in the embodiment of this application is introduced.

[0060] With the rapid development of digital design technology, the automatic generation of 3D clothing design drawings has become an important direction for the intelligentization of the clothing industry. Through computer-aided design, designers can more efficiently realize their creative expression, shorten the design cycle, and meet personalized needs.

[0061] Related technologies typically employ image prediction techniques based on convolutional neural networks (CNNs) combined with principal component analysis (PCA) to simplify clothing panel data, or employ differentiable simulation optimization techniques to restore clothing material properties and pattern details. These methods improve the automation and refinement of design to a certain extent, but often rely on complex parameter settings and image processing processes, lack intuitive and diverse user interaction methods, and struggle to meet the convenience and flexibility users require when expressing their design intent. This results in a poor user experience and limited applicability. Furthermore, existing technologies have yet to achieve direct and efficient interaction with natural language, leading to an information gap between design intent and generated results.

[0062] Therefore, in the technology of automatic generation of three-dimensional clothing design drawings, the complexity of user interaction and the inconvenience of operation, as well as the poor interaction effect between natural language and design models, have become problems that need to be solved urgently.

[0063] Based on the problems existing in related technologies, this application provides a three-dimensional clothing design drawing generation method. By obtaining structured text information describing the color, style, material and details of the clothing as the input condition of the three-dimensional clothing design drawing generation model, the three-dimensional clothing design drawing generation model is used to predict and process the text information to generate a three-dimensional clothing design drawing that meets the user's design intention.

[0064] The structured text information in this application includes details related to the 3D clothing design drawings, such as patterns, collars, sleeves, hems, and trouser legs. This enhances the convenience and flexibility of users in expressing their design needs, effectively overcoming the shortcomings of traditional technologies that rely on complex parameter input and a single interactive method. Furthermore, the 3D clothing design generation model is trained based on a large sample of 3D design drawings and corresponding text information. This model enables efficient and accurate mapping between natural language and 3D clothing design drawings, effectively narrowing the information gap between design intent and generated results.

[0065] In summary, the three-dimensional clothing design drawing generation method provided in this application improves the user-friendliness, interaction efficiency and accuracy of design expression of the three-dimensional clothing design drawing generation process through a text-driven generation mechanism and structured detail control, thereby effectively solving the technical problems in the background technology that urgently need to be broken through, such as the complexity of user interaction, inconvenience of operation and low efficiency of natural language interaction.

[0066] Next, some specific embodiments and drawings are used to describe in detail how the present application solves the problems of complex user interaction and inconvenient operation in existing three-dimensional clothing design drawing automatic generation technology, as well as poor interaction between natural language and design models.

[0067] Figure 1 This is a flow chart of a method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application. Figure 1 As shown, the method for generating a three-dimensional clothing design drawing provided in the embodiment of the present application specifically includes S101 and S102, and S101 and S102 are described in detail below.

[0068] It should be noted that the execution entity of the three-dimensional clothing design drawing generation method provided in the embodiment of the present application can be a controller in a terminal device, and the terminal device can be a mobile phone, a computer, etc.

[0069] S101: Obtain first text information.

[0070] The first text information is used to describe the color information, style information, material information and detail information of the desired first three-dimensional clothing design drawing.

[0071] It should be noted that the first text information uses natural language to describe key attributes of the first three-dimensional clothing design drawing that the user expects to generate, including color information, style information, material information and detail information.

[0072] The color information is used to describe the main color, auxiliary color and distribution characteristics of the first three-dimensional clothing design drawing.

[0073] Specifically, color information supports hierarchical descriptions, including primary, secondary, and accent colors. This facilitates the detailed color distribution of each part of a 3D clothing design. Primary colors are used for the main body of the 3D clothing design, secondary colors are used for decorative patterns within the 3D clothing design, and accent colors are used for details like buttons and zippers.

[0074] The style information is used to describe the overall style and structural features of the first three-dimensional clothing design drawing.

[0075] Specifically, the style information includes at least one of clothing category, clothing silhouette, and design style.

[0076] The clothing category is used to indicate the basic type of the first three-dimensional clothing design drawing, such as dress, shirt, suit, jacket, pants, etc., to ensure that the three-dimensional clothing design drawing generation model can generate corresponding structural features and design styles for different clothing categories.

[0077] The clothing outline is used to indicate the overall outline shape of the first three-dimensional clothing design drawing, such as slim, loose, straight, fluffy, etc.

[0078] The design style is used to indicate the style attributes of the first three-dimensional clothing design drawing, such as casual, formal, sports, retro, modern, etc., to provide the style guidance of the first three-dimensional clothing design drawing for the three-dimensional clothing design drawing generation model, and assist the first three-dimensional clothing design drawing in generating a first three-dimensional clothing design drawing that conforms to a specific design style.

[0079] The material information is used to describe the type of fabric used in the first three-dimensional clothing design drawing and its physical properties.

[0080] Specifically, the material information includes at least one of fabric category, physical properties, and texture characteristics.

[0081] The fabric category is used to indicate the fabric used in the first three-dimensional clothing design drawing, such as cotton, silk, wool, denim, leather, lace, etc., to ensure that the three-dimensional clothing design drawing generation model can generate the first three-dimensional clothing design drawing with corresponding surface texture and form expression according to different material characteristics.

[0082] Physical properties are used to indicate the physical characteristics of fabrics such as softness, elasticity, thickness, breathability, gloss, etc.

[0083] Texture features are used to indicate the texture details of the fabric surface, such as fineness, roughness, fuzziness, stripes, plaids, etc., so that the first three-dimensional clothing design drawing output by the three-dimensional clothing design drawing generation model can show a realistic visual effect.

[0084] The detail information is used to indicate at least one of a pattern, a collar type, a sleeve type, a skirt hem, and a trouser leg related to the first three-dimensional clothing design drawing.

[0085] It should be noted that the detail information is used to refine and enrich the key design elements in the first three-dimensional clothing design drawing.

[0086] The pattern is used to indicate the decorative graphics contained in the surface of the first three-dimensional clothing design drawing, such as prints, embroidery, geometric textures, stripes, polka dots, flowers, etc. The type, color, size, arrangement, etc. of the pattern can all be described through detailed information.

[0087] The collar type is used to indicate the shape and style of the collar of the first three-dimensional clothing design drawing, such as a stand-up collar, a lapel collar, a round collar, a V-neck, a boat neck, etc.

[0088] The sleeve type is used to indicate the shape and length of the sleeves of the first three-dimensional clothing design drawing, such as sleeveless, short sleeves, three-quarter sleeves, long sleeves, lantern sleeves, puff sleeves, etc.

[0089] When the first three-dimensional clothing design drawing is a skirt, the hem is used to indicate the hem shape of the first three-dimensional clothing design drawing, such as a straight skirt, an umbrella skirt, a fishtail skirt, an A-line skirt, etc.

[0090] When the first three-dimensional clothing design drawing is pants, the trouser legs are used to indicate the design shape of the bottom of the trouser legs of the first three-dimensional clothing design drawing, such as narrow-leg pants, bell-bottom pants, straight-leg pants, and tapered pants.

[0091] In this embodiment, the first text information input by the user uses natural language to describe in detail the color, style, material, and detailed attributes of the first three-dimensional clothing design that the user desires. This facilitates the three-dimensional clothing design generation model to accurately capture the user's design intent, laying a solid foundation for the subsequent accurate generation of the first three-dimensional clothing design. The three-dimensional clothing design generation method of this application supports multi-dimensional and multi-level design requirement input, greatly improving the convenience and flexibility of user interaction and overcoming the limitations of traditional complex parameter input methods.

[0092] S102: performing prediction processing on the first text information based on a three-dimensional clothing design drawing generation model to obtain and output a first three-dimensional clothing design drawing.

[0093] The three-dimensional clothing design drawing generation model is obtained by training the three-dimensional clothing design drawing samples and the second text information corresponding to the three-dimensional clothing design drawing samples.

[0094] It should be noted that the 3D clothing design drawing generation model is obtained by using a large number of 3D clothing design drawing samples and the second text information corresponding to the 3D clothing design drawing samples to supervise the learning of the initial 3D clothing design drawing generation model.

[0095] During the training process of the initial three-dimensional clothing design drawing generation model, the initial three-dimensional clothing design drawing generation model obtains the three-dimensional clothing design drawing generation model by continuously learning the complex relationship between the second text information and the second text information and the corresponding three-dimensional clothing design drawing samples, and continuously optimizing the loss function.

[0096] The second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design sample.

[0097] It should be noted that the method of using the second text information to describe the three-dimensional clothing design drawing sample is similar to the method of using the first text information to describe the first three-dimensional clothing design drawing, which will not be described in detail in this embodiment.

[0098] In this embodiment, the first text information is used as input to the 3D clothing design generation model, which then performs efficient predictive processing on the first text information to generate the first 3D clothing design, achieving a precise mapping from text description to 3D clothing design. This not only enhances the efficiency of interaction between natural language and the 3D clothing design generation model, narrowing the information gap between design intent and generated results, but also meticulously restores the color, style, material, and detailed features of the garment, significantly improving the authenticity and professionalism of the first 3D clothing design generated by the 3D clothing design generation model, thus meeting users' demand for high-quality, personalized 3D clothing design.

[0099] It should be noted that the prediction processing includes encoding processing of the first text information, cross-attention processing of the starting token and the first text information after encoding (first text guide vector), and decoding processing of the target token sequence obtained by the cross-attention processing.

[0100] The embodiment of the present application provides a three-dimensional clothing design drawing generation method, which uses first text information as input to a three-dimensional clothing design drawing generation model. The first text information uses natural language to describe the color information, style information, material information and detail information of the first three-dimensional clothing design drawing that the user expects to obtain, and the three-dimensional clothing design drawing generation model predicts and processes the first text information to obtain the corresponding first three-dimensional clothing design drawing, thereby achieving efficient conversion from natural language to three-dimensional clothing design drawings. The three-dimensional clothing design drawing generation model uses text information expressed in natural language as input, which significantly improves the convenience and flexibility of users in expressing their design intentions. At the same time, the first text information is used to accurately express the color, style, material and details of the clothing, effectively narrowing the information gap between the design intention and the generated results, and improving the authenticity and professionalism of the three-dimensional clothing design drawings.

[0101] In the above embodiment, the terminal device needs to perform predictive processing on the first text information based on the 3D clothing design drawing generation model to obtain and output the first 3D clothing design drawing. Next, the specific process of the terminal device performing predictive processing on the first text information based on the 3D clothing design drawing generation model to obtain and output the first 3D clothing design drawing will be described in detail.

[0102] Figure 2 This is a flow chart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application. Figure 2 As shown, in a possible embodiment, S102 can be implemented through S1021 to S1023, and S1021 to S1023 are described in detail below.

[0103] S1021: Process the first text information to obtain a first text guidance vector.

[0104] The first text guidance vector is a semantic representation of the first text information, and can accurately reflect the semantic features of the color information, style information, material information and detail information of the first three-dimensional clothing design drawing that the user expects to obtain.

[0105] It should be noted that the terminal device may use a contrastive language-image pre-training (CLIP) model to encode the first text information to obtain a first text guidance vector.

[0106] Specifically, the method of encoding the first text information through the CLIP model to obtain the first text guide vector is an existing method, which will not be described in detail in this embodiment.

[0107] In another embodiment, the terminal device may preprocess and encode the first text information input by the user through a natural language processing module to obtain a first text guidance vector.

[0108] Specifically, first, a tokenizer is used to split the first text information into multiple tokens; second, a pre-trained language model (such as BERT, GPT, etc.) is used to map each token to a shared high-dimensional embedding space; finally, a text encoder (Transformer encoding layer, recurrent neural network, etc.) is used to perform context modeling on the embedding vector, fuse semantic information, and output the first text guidance vector.

[0109] In this embodiment, by preprocessing and encoding the first text information, a high-quality first text guidance vector is generated. The first text guidance vector can accurately capture the multidimensional semantic information of the user's design intent, providing a precise semantic foundation for the subsequent generation of the first three-dimensional clothing design drawing. At the same time, this effectively improves the expressiveness and semantic consistency of the text information, enhances the correlation between natural language and the three-dimensional clothing design drawing generation model, and significantly improves the three-dimensional clothing design drawing generation model's understanding depth and response accuracy of design requirements, thereby ensuring the accuracy and personalization of the generated first three-dimensional clothing design drawing.

[0110] S1022. Based on the three-dimensional clothing design drawing generation model, cross-attention processing is performed on the starting token and the first text guide vector to obtain a target token sequence.

[0111] Among them, the starting token is a fixed starting symbol.

[0112] It should be noted that the terminal device inputs the starting token and the first text guidance vector into the cross-attention layer in the Transformer architecture. The cross-attention layer focuses on the key semantic features in the first text guidance vector by calculating the attention weight between the starting token and the first text guidance vector, and gradually generates a target token sequence that conforms to the semantics of the first text guidance vector.

[0113] The target token sequence is a set of encoding sequences generated by combining the start token and the first text guide vector through a cross-attention mechanism. The target token sequence is used to construct the first 3D clothing design image.

[0114] It should be noted that the target token sequence encodes the construction instructions or feature representations used in the generation of the first 3D clothing design drawing, using tokens as units. Generating this target token sequence enables the effective mapping of natural language descriptions to 3D clothing designs, promoting the deep integration of textual semantic information and 3D model features, and laying the foundation for accurate and high-quality generation of the first 3D clothing design drawing.

[0115] In this embodiment, through cross-attention processing, the first text guidance vector effectively guides and constrains the generation process of the first three-dimensional clothing design drawing, ensuring that the generated first three-dimensional clothing design drawing accurately matches user requirements in color, style, material and details, and improves the collaborative efficiency of the first text information and the three-dimensional clothing design drawing generation model, significantly narrowing the gap between design intent and generation results, thereby improving the generation quality and expressiveness of the first three-dimensional clothing design drawing generated by the three-dimensional clothing design drawing generation model.

[0116] S1023. Based on the target token sequence, obtain and output a first three-dimensional clothing design drawing.

[0117] After obtaining the target token sequence, the terminal device inputs the target token sequence into a decoding module, and the decoding module converts the target token sequence into a specific representation of the first three-dimensional clothing design drawing.

[0118] Specifically, the decoding module can use a convolutional network, a graph neural network or a voxel-based, point cloud-based three-dimensional reconstruction algorithm to gradually restore the spatial structure, color distribution, material texture and detailed form of the first three-dimensional clothing design drawing, and finally generate a first three-dimensional clothing design drawing that is highly consistent with the description of the first text information.

[0119] This embodiment realizes the precise conversion from the semantic sequence to the first three-dimensional clothing design drawing. The generated first three-dimensional clothing design drawing not only meets the various design requirements described in the first text information, but also has good visual performance and practicality, which significantly improves the efficiency and quality of the generation of three-dimensional clothing design drawings.

[0120] In an embodiment of the present application, a first text guidance vector is obtained by efficiently encoding the first text information, and a cross-attention mechanism is used to achieve deep fusion of the first text guidance vector and the three-dimensional clothing design drawing generation model, thereby generating a target token sequence for constructing the first three-dimensional clothing design drawing. Finally, the three-dimensional clothing design drawing is accurately restored based on the target token sequence, which significantly improves the ability of natural language to guide the three-dimensional clothing design drawing generation process, and effectively solves the problems of poor interaction between natural language and design models and mismatch between design intent and generation results in the prior art.

[0121] In the above embodiment, the terminal device needs to perform cross-attention processing on the start token and the first text guidance vector to obtain the target token sequence. Next, the specific process of the terminal device performing cross-attention processing on the start token and the first text guidance vector to obtain the target token sequence is described in detail.

[0122] In a possible embodiment, S1022 may be implemented by Sa1 to Sa4, which are described in detail below.

[0123] Sa1. Perform cross-attention processing on the starting token and the first text guide vector to obtain the first token.

[0124] The terminal device inputs the preset starting token and the first text guidance vector into the cross-attention layer. The cross-attention layer uses the starting token as the query (Query), the first text guidance vector as the (Key) and the value (Value), calculates the attention weight, and thus focuses on the key information in the first text guidance vector and generates a first token associated with the semantics of the first text guidance vector.

[0125] The first token is the initial generation result of the target token sequence, reflecting the first step response to the first text guidance vector.

[0126] Through this step, this embodiment effectively activates the guiding role of the first text guidance vector in the generation of the target token sequence, enabling the three-dimensional clothing design drawing generation model to accurately guide the generation of subsequent tokens based on the semantic features of the first text guidance vector, thereby ensuring the semantic coherence and accuracy of the generated target token sequence.

[0127] Sa2. Concatenate the start token and the first token to obtain the first token sequence.

[0128] The preset start token and the first token are concatenated in the sequence dimension to form a first token sequence of length 2. The first token sequence serves as the initial context for the generation process of the target token sequence and provides input for the next step of cross-attention processing.

[0129] Sa3. Set the first token sequence as the starting token and repeat the above cross-attention processing steps until the ending token is obtained.

[0130] After the terminal device sets the first token sequence as the start token, it repeats the method steps shown in Sa1 above until the end token appears.

[0131] It should be noted that the terminal device takes the first token sequence as the new starting token, and inputs the new starting token and the first text guidance vector into the cross-attention layer. The cross-attention layer performs cross-attention processing on the new starting token and the first text guidance vector to obtain a new first token.

[0132] The cross-attention layer performs cross-attention processing on the new starting token and the first text guide vector. The specific process of obtaining the new first token is similar to the specific process of obtaining the first token by Sa1 mentioned above, which will not be repeated in this embodiment.

[0133] The end token is a fixed end symbol and is used to mark the end of the target token sequence generation.

[0134] In this embodiment, the three-dimensional clothing design drawing generation model can dynamically capture the relationship between the semantics of the first text guidance vector and the generated content through cyclic recursive cross-attention processing, realize the gradual refinement and expression of complex design information, and ensure that the final generated token sequence is complete and semantically rich.

[0135] Sa4. Concatenate the end token and the start token corresponding to the end token to obtain the target token sequence.

[0136] After generating the end token, the system concatenates the end token with the corresponding start token in the sequence dimension to form a complete target token sequence. The target token sequence includes all tokens generated between the start token and the end token.

[0137] For example, the first text message entered by the user is: "Design a red dress with lace and puff sleeves." Among them, the starting token is " <start>", the termination token is " <end>”.

[0138] First, the cross attention layer is <start>" is used as the query, combined with the first text guide vector obtained by encoding the first text information, the attention weight is calculated, and the key information with the semantics of "red", "dress", "lace", and "puff sleeves" in the first text guide vector is focused on, and the first token is generated. The first token is "dress", and " <start>" and "dress" are concatenated into the first token sequence, and the first token sequence is " <start>dress".

[0139] Secondly, the first token sequence " <start>dress" as the new starting token, combined with the first text guide vector, continue the cross attention process to generate the next token, the next token is "red", and then splice to form the next token sequence, the next token sequence is " <start>dress red"; then, the next token sequence is " <start>dress red" as the new starting token, repeat the above steps several times until the generated token is " <end>"; Among them, in the token sequence" <start>dress red lace puff sleeves" as the new starting token, the generated token is " <end>”.

[0140] Finally, the token sequence " <start>dress red lace puff sleeves" and termination token" <end>"Splice together to get the target token sequence, the target token sequence is" <start>dress red lace puffsleeves <end>”.

[0141] In this embodiment of the present application, by recursively inputting the starting token and the first text guidance vector into the cross-attention layer, a complete target token sequence is gradually generated, achieving efficient conversion from natural language to 3D clothing design generation instructions. This helps ensure the semantic coherence and accuracy of the generated target token sequence, improves the ability of text information to guide the 3D clothing design generation process, effectively solves the mismatch between design intent and generated results in the prior art, and significantly improves the generation quality and user satisfaction of 3D clothing design drawings.

[0142] In the above embodiment, the terminal device needs to perform cross-attention processing on the start token and the first text guidance vector to obtain the first token. Next, the specific process of the terminal device performing cross-attention processing on the start token and the first text guidance vector to obtain the first token is described in detail.

[0143] In a possible embodiment, Sa1 may be implemented by Sa11 to Sa13, which are described in detail below.

[0144] Sa11. Based on the position encoder, parameter encoder and value encoder, the starting token is processed respectively to obtain the position latent vector, parameter latent vector and value latent vector corresponding to the starting token.

[0145] It should be noted that by processing the start token through the position encoder, the position latent vector corresponding to the start token can be obtained; by processing the start token through the parameter encoder, the parameter latent vector corresponding to the start token can be obtained; and by processing the start token through the value encoder, the value latent vector corresponding to the start token can be obtained.

[0146] Among them, the position encoder, parameter encoder and value encoder are trained through three-dimensional clothing design samples.

[0147] It should be noted that the position encoder is trained to understand and represent the position of each part of a 3D clothing design within a sample. For example, the positional relationship and relative position of different parts of a 3D clothing design (such as the neckline, cuffs, and hem) in 3D space. The position encoder learns from the sample 3D clothing design how to convert features at different spatial locations into a position latent vector. This position latent vector is used to accurately indicate the spatial position of each part of the 3D clothing design during the subsequent generation of the 3D clothing design.

[0148] The parameter encoder is trained to understand and represent parameters related to a sample 3D clothing design, such as color, material, and style. Using the design information contained in the sample, the parameter encoder learns how to convert these attributes into parameter latent vectors. These latent vectors are then used to accurately represent the parameter information for each component of the 3D clothing design during the subsequent generation of the 3D clothing design.

[0149] The value encoder is trained to understand and represent numerical values ​​associated with 3D clothing design samples, such as dimensions, proportions, and angles. Based on the specific numerical data in the 3D clothing design samples, the value encoder generates corresponding value latent vectors. For example, the value encoder learns how to convert numerical information such as "shoulder width: 40cm" or "waist circumference: 70cm" into a format that a computer can understand.

[0150] The position latent vector corresponding to the starting token is used to indicate the position of the starting token on the first three-dimensional clothing design drawing.

[0151] It should be noted that the position latent vector is a high-dimensional vector. The position latent vector of the starting token represents the spatial position information of the starting token in the 3D clothing design image. Each starting token will generate a corresponding position latent vector based on its position characteristics in the 3D clothing design image.

[0152] The positional features of the starting token within the 3D clothing design include spatial coordinates and relative positional relationships. Determining the positional latent vector corresponding to the starting token helps the 3D clothing design generation model accurately place the various components of the first 3D clothing design during the generation process, ensuring that the spatial layout of each design element aligns with user requirements.

[0153] Among them, the parameter latent vector corresponding to the starting token is used to indicate the parameter type of the starting token.

[0154] It should be noted that the parameter latent vector is also a high-dimensional vector. The parameter latent vector of the starting token represents the design information of the starting token, including its color, style, material, and detail information. By determining the parameter latent vector corresponding to the starting token, the 3D clothing design generation model can accurately understand the design requirements and attributes of each part in the first 3D clothing design during the generation process, ensuring that the subsequently generated first 3D clothing design accurately reflects the user's design intent.

[0155] Among them, the value latent vector corresponding to the starting token is used to indicate the numerical information corresponding to the starting token.

[0156] It's important to note that the value latent vector of the start token represents the numerical information associated with the start token. Unlike parameter latent vectors, value latent vectors directly correspond to specific numerical values ​​in the 3D clothing design. Value latent vectors can include at least one of the following numerically related information: size, scale, angle, etc.

[0157] For example, a value latent vector can include size information, such as "shoulder width: 40cm" and "trouser length: 100cm." A value latent vector can also include proportion information, such as "chest-to-waist ratio." A value latent vector can also include angle information, such as whether the sleeves of certain designs are at a specific angle (45-degree cuffs).

[0158] By determining the value latent vector corresponding to the starting token, the 3D clothing design drawing generation model can generate 3D clothing design drawings that meet actual wearing needs according to specific size, proportion or angle requirements.

[0159] Sa12. Concatenate the starting token and the position latent vector, parameter latent vector, and value latent vector corresponding to the starting token to obtain the second token sequence.

[0160] The terminal device concatenates the starting token and its corresponding positional latent vector, parameter latent vector, and value latent vector to form a second token sequence. This second token sequence incorporates the starting token's positional information, parameter information, and numerical information. This allows the 3D clothing design generation model to simultaneously reference the starting token's positional information, parameter information, and numerical information when generating the first token, ensuring that the subsequently generated first 3D clothing design better aligns with the user's design intent.

[0161] Sa13. Perform cross-attention processing on the second token sequence and the first text guide vector to obtain the first token.

[0162] It should be noted that the method of performing cross-attention processing on the second token sequence and the first text guidance vector to obtain the first token is similar to the method steps shown in Sa1 above, in which the starting token and the first text guidance vector are cross-attention processed to obtain the first token. This embodiment will not go into details about this.

[0163] In an embodiment of the present application, by performing cross-attention processing on the starting token and the first text guide vector, and combining the position latent vector, parameter latent vector and value latent vector generated by the position encoder, parameter encoder and value encoder based on the starting token, the spatial position, attribute information and numerical information of each starting token are accurately encoded and fused, thereby effectively enhancing the ability of the three-dimensional clothing design drawing generation model to control the position, attributes and details of each part in the three-dimensional clothing design drawing, and ensuring that the generated target token sequence can accurately reflect the user's design intention.

[0164] In the above embodiment, the terminal device needs to perform prediction processing on the first text information based on the 3D clothing design drawing generation model to obtain and output the first 3D clothing design drawing. Next, the specific process of obtaining the 3D clothing design drawing generation model is described in detail.

[0165] Figure 3 This is a flow chart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application. Figure 3 As shown, in a possible embodiment, before the method step shown in S101, the three-dimensional clothing design drawing generation method further includes S301 and S302, and S301 and S302 are described in detail below.

[0166] S301: Obtain a training sample set.

[0167] The training sample set includes a plurality of training samples, and the training samples are three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples.

[0168] It should be noted that the second text information corresponding to each 3D clothing design sample is a descriptive text of the 3D clothing design sample, and describes the color information, style information, material information, and detail information of the corresponding 3D clothing design sample.

[0169] The second text information may be manually input by a designer in advance, or may be extracted through image recognition technology, which is not specifically limited in this embodiment.

[0170] S302: Based on multiple training samples, the initial three-dimensional clothing design drawing generation model is trained to obtain a three-dimensional clothing design drawing generation model.

[0171] The initial 3D clothing design generation model is a pre-defined model structure, which may be based on a neural network, a generative adversarial network, or other deep learning model. This model needs to be gradually optimized through training to enable it to generate 3D clothing designs that meet the textual requirements based on the input text.

[0172] In the process of training the initial three-dimensional clothing design drawing generation model, the initial three-dimensional clothing design drawing generation model receives multiple training samples and adjusts parameters through algorithms (such as back propagation, gradient descent, etc.) so that the output of the initial three-dimensional clothing design drawing generation model gradually approaches the three-dimensional clothing design drawing indicated by the training samples, thereby obtaining a three-dimensional clothing design drawing generation model.

[0173] In the embodiments of the present application, a rich and diverse training sample set is constructed by acquiring a large number of high-quality 3D clothing design sample images and the corresponding second text information for each 3D clothing design sample image. This provides sufficient data support for the training of the initial 3D clothing design image generation model, thereby obtaining the 3D clothing design image generation model. Upon receiving new natural language text information, the 3D clothing design image generation model can quickly and accurately generate high-quality 3D clothing design images.

[0174] In the above embodiment, the terminal device needs to train the initial 3D clothing design drawing generation model based on multiple training samples to obtain the 3D clothing design drawing generation model. Next, the specific process of training the initial 3D clothing design drawing generation model based on multiple training samples to obtain the 3D clothing design drawing generation model will be described in detail.

[0175] Figure 4 This is a flow chart of another method for generating a three-dimensional clothing design drawing provided in an embodiment of the present application. Figure 4 As shown, in a possible embodiment, the method steps shown in S302 can be implemented through S3021 to S3024, and S3021 to S3024 are described in detail below.

[0176] S3021. For any training sample among the multiple training samples, obtain a second text guidance vector based on second text information corresponding to the three-dimensional clothing design drawing sample.

[0177] Among them, for any training sample, the method of obtaining the second text guidance vector based on the second text information corresponding to the three-dimensional clothing design drawing sample is similar to the method of processing the first text information in the above S1021 to obtain the first text guidance vector, and this embodiment will not repeat it again.

[0178] S3022. Perform cross-attention processing on the starting token and the second text guide vector to obtain a third token sequence corresponding to the three-dimensional clothing design sample.

[0179] Among them, the terminal device performs cross-attention processing on the starting token and the second text guidance vector to obtain a third token sequence corresponding to the three-dimensional clothing design drawing sample, which is similar to the method of performing cross-attention processing on the starting token and the first text guidance vector in the above S1022 to obtain the target token sequence, and this embodiment will not repeat it.

[0180] S3023. Determine a fourth token sequence corresponding to the three-dimensional clothing design sample using a preset rule.

[0181] The preset rules may be standard templates for clothing design, standardized expressions of design drawings, etc., which are not specifically limited in this embodiment.

[0182] It should be noted that by determining the fourth token sequence corresponding to the three-dimensional clothing design drawing sample through preset rules, a standardized and unified reference label can be provided for each three-dimensional clothing design drawing sample, which simplifies the sample labeling in the initial three-dimensional clothing design drawing generation model training process, ensures the consistency between the samples learned by the initial three-dimensional clothing design drawing generation model and the actual application, and makes the subsequently obtained three-dimensional clothing design drawing generation model more reliable.

[0183] S3024. Construct a loss function based on the difference between the third token sequence and the fourth token sequence, and update the initial three-dimensional clothing design drawing generation model based on the loss function to obtain the three-dimensional clothing design drawing generation model.

[0184] For the same three-dimensional clothing design drawing sample, the third token sequence corresponding to the three-dimensional clothing design drawing sample is obtained by the initial three-dimensional clothing design drawing generation model through the cross-attention mechanism, and the third token sequence is related to the second text guidance vector corresponding to the three-dimensional clothing design drawing sample; the fourth token sequence corresponding to the three-dimensional clothing design drawing sample is a token sequence obtained by processing the three-dimensional clothing design drawing sample through preset rules, and the fourth token sequence represents the standard output of the three-dimensional clothing design drawing.

[0185] The loss function is used to quantify the difference between the third and fourth token sequences. By minimizing the loss function, the initial 3D clothing design generation model can be continuously optimized, making the 3D clothing design output by the initial 3D clothing design generation model closer to the 3D clothing design sample.

[0186] In an embodiment of the present application, the parameters of the initial 3D clothing design generation model are continuously optimized by calculating the difference between the third and fourth token sequences and constructing a loss function based on this difference, thereby obtaining a 3D clothing design generation model. The 3D clothing design generation model can accurately extract design elements from text information described in natural language and generate high-quality 3D clothing design drawings that meet the user's design intent, effectively resolving the problem of mismatch between generated results and design intent in existing technologies.

[0187] In the above embodiment, the terminal device needs to use a preset rule to determine the fourth token sequence corresponding to the three-dimensional clothing design sample. Next, the specific process of the terminal device using the preset rule to determine the fourth token sequence corresponding to the three-dimensional clothing design sample is described in detail.

[0188] In a possible embodiment, the method steps shown in S3023 may be implemented by Sc1 and Sc2, which are described in detail below.

[0189] Sc1. Determine the rotation quaternions and translation vectors corresponding to the multiple polygonal panels corresponding to the three-dimensional clothing design sample, as well as the edge information and stitching information corresponding to the multiple edges of each polygonal panel.

[0190] It should be noted that to facilitate computer processing of 3D clothing design samples, they are typically segmented into multiple 2D polygonal panels (e.g., hexagons, quadrilaterals, triangles, etc.). Each polygonal panel represents a component of the 3D clothing design sample, and multiple polygonal panels connected together form a complete 3D clothing design sample.

[0191] For example, a 3D clothing design sample might be a skirt. The skirt can be divided into multiple parts, such as the skirt body, hem, and waistband. Each part can be represented by at least one polygonal panel. Each polygonal panel can be an unfolded view of a certain part of the skirt, depicting the fabric pattern, size, shape, etc.

[0192] The rotation quaternion is used to indicate the rotation angle of the polygon panel relative to the global coordinate system.

[0193] It's important to note that the global coordinate system is a coordinate system used to describe the position of objects or points in space. It provides a fixed, unified spatial positioning standard for all objects. Regardless of how an object rotates or translates, the global coordinate system accurately determines its spatial position.

[0194] The translation vector is used to indicate the position offset of the polygon panel relative to the global coordinate system.

[0195] The edge information is used to indicate the start point, end point, and Bezier curve control point of the edge.

[0196] It should be noted that the Bezier curve control points are used to describe the curve shape of the edge.

[0197] Based on the start and end points of each edge, as well as the Bezier curve control points of each edge, the position of each edge in space can be precisely determined.

[0198] The stitching information is used to indicate whether to stitch the edge and the position of the target edge to be stitched.

[0199] It should be noted that each polygonal panel has corresponding identification information to facilitate accurate identification and location of each polygonal panel. The identification information of the polygonal panel includes, but is not limited to, the panel type, size, design attributes, etc. The identification information of the polygonal panel can be used to distinguish different types of polygonal panels.

[0200] In addition, each edge of each polygonal panel also has its own corresponding identification information. Each edge's identification information includes the coordinates of the edge's starting and ending points, as well as the coordinates of the edge's Bezier curve control points. This identification information allows for accurate representation of each edge's edge information and stitching information.

[0201] Sc2. Concatenate the rotation quaternions and translation vectors corresponding to the multiple polygonal panels, as well as the edge information and stitching information corresponding to the multiple edges of each polygonal panel, to obtain a fourth token sequence.

[0202] It should be noted that the fourth token sequence concatenates the geometric, structural, and stitching information of all polygonal panels to form a comprehensive representation that includes all necessary design features.

[0203] In an embodiment of the present application, a fourth token sequence is obtained by splicing the rotation quaternions, translation vectors, edge information and stitching information corresponding to all polygonal panels. The fourth token sequence centrally represents the geometric information, structural information and stitching information of all polygonal panels, providing a comprehensive description of the three-dimensional clothing design drawing sample.

[0204] Figure 5 This is a schematic diagram of the structure of a three-dimensional clothing design drawing generation device provided in an embodiment of the present application. Figure 5 As shown, the three-dimensional clothing design drawing generating device provided by this embodiment includes an acquisition module 501 and an output module 502.

[0205] Among them, the acquisition module 501 is used to obtain the first text information, and the first text information is used to describe the color information, style information, material information and detail information of the first three-dimensional clothing design drawing expected to be obtained; wherein the detail information is used to indicate at least one of the pattern, collar type, sleeve type, skirt hem, and trouser leg related to the first three-dimensional clothing design drawing.

[0206] Output module 502 is used to predict and process the first text information based on a three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing; wherein, the three-dimensional clothing design drawing generation model is obtained by training three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples, and the second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design drawing samples.

[0207] It should be understood that the execution of the above corresponding processes by each module has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here.

[0208] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 As shown, the electronic device 600 provided in this embodiment includes: a memory 601 and a processor 602.

[0209] The memory 601 may be an independent physical unit and may be connected to the processor 602 via a bus 603. The memory 601 and the processor 602 may also be integrated together and implemented via hardware. The memory 601 is used to store program instructions, and the processor 602 calls the program instructions to execute the operations performed by the terminal device in any of the above method embodiments.

[0210] Optionally, when part or all of the methods of the above embodiments are implemented by software, the electronic device 600 may also include only a processor 602. The memory 601 for storing programs is located outside the electronic device 600, and the processor 602 is connected to the memory via circuits / wires to read and execute the programs stored in the memory. The processor 602 may be a central processing unit (CPU), a network processor (NP), or a combination of a CPU and an NP. The processor 602 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0211] The memory 601 may include a volatile memory, such as a random-access memory (RAM); the memory may also include a non-volatile memory, such as a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD); the memory may also include a combination of the above types of memory.

[0212] Illustratively, the present application provides a chip comprising: an interface circuit and a logic circuit, wherein the interface circuit is used to receive signals from other chips outside the chip and transmit them to the logic circuit, or to send signals from the logic circuit to other chips outside the chip, and the logic circuit is used to execute the operations performed by the terminal device in the above method embodiment.

[0213] Illustratively, the present application provides a computer-readable storage medium having computer program instructions stored thereon. The computer program instructions are executed by a processor of an electronic device so that the electronic device performs the operations performed by the terminal device in the above method embodiment.

[0214] Illustratively, the present application provides a computer program product, which, when executed on an electronic device, enables the electronic device to execute the operations executed by the terminal device in the above method embodiment.

[0215] The foregoing description is intended only to provide specific embodiments of the present application, which will enable those skilled in the art to understand and implement the present application. Various modifications to these embodiments will be readily apparent to those 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 application. Therefore, the present application is not limited to the embodiments described herein, but is intended to be construed in the broadest manner consistent with the principles and novel features disclosed herein.< / end> < / start> < / end> < / start> < / end> < / start> < / end> < / start> < / start> < / start> < / start> < / start> < / start> < / end> < / start>

Claims

1. A method for generating a three-dimensional clothing design drawing, characterized in that: The method comprises: Obtaining first text information, the first text information being used to describe color information, style information, material information, and detail information of a desired first three-dimensional clothing design drawing; wherein the detail information is used to indicate at least one of a pattern, a collar type, a sleeve type, a skirt hem, and a trouser leg related to the first three-dimensional clothing design drawing; The first text information is predicted and processed based on a three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing; wherein the three-dimensional clothing design drawing generation model is obtained by training three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples, and the second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design drawing samples.

2. The method according to claim 1, characterized in that The predictive processing of the first text information based on the three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing includes: Processing the first text information to obtain a first text guidance vector; Based on the three-dimensional clothing design drawing generation model, cross-attention processing is performed on the starting word and the first text guide vector to obtain a target word sequence; Based on the target word sequence, the first three-dimensional clothing design drawing is obtained and output.

3. The method according to claim 2, characterized in that The cross-attention processing is performed on the starting word and the first text guide vector to obtain a target word sequence, including: performing a cross-attention process on the starting word-unit and the first text guide vector to obtain a first word-unit; concatenating the starting word and the first word to obtain a first word sequence; Setting the first word-gram sequence as the starting word-gram, and repeatedly performing the above cross-attention processing steps until an ending word-gram is obtained; The end word and the start word corresponding to the end word are concatenated to obtain the target word sequence.

4. The method according to claim 3, characterized in that The performing cross attention processing on the starting word unit and the first text guide vector to obtain a first word unit includes: The starting word is processed based on a position encoder, a parameter encoder, and a value encoder respectively to obtain a position latent vector, a parameter latent vector, and a value latent vector corresponding to the starting word; wherein the position encoder, the parameter encoder, and the value encoder are obtained by training the three-dimensional clothing design drawing sample; the position latent vector corresponding to the starting word is used to indicate the position of the starting word on the first three-dimensional clothing design drawing; the parameter latent vector corresponding to the starting word is used to indicate the parameter type of the starting word; and the value latent vector corresponding to the starting word is used to indicate the numerical information corresponding to the starting word; Concatenating the starting word, and the position latent vector, the parameter latent vector, and the value latent vector corresponding to the starting word to obtain a second word sequence; Cross-attention processing is performed on the second word-gram sequence and the first text guide vector to obtain the first word-gram.

5. The method according to claim 1, wherein Before obtaining the first text information, the method further includes: Acquire a training sample set, wherein the training sample set includes a plurality of training samples, and the training samples are the three-dimensional clothing design drawing samples and the second text information corresponding to the three-dimensional clothing design drawing samples; Based on the multiple training samples, the initial three-dimensional clothing design drawing generation model is trained to obtain the three-dimensional clothing design drawing generation model.

6. The method according to claim 5, characterized in that The initial three-dimensional clothing design drawing generation model is trained based on the multiple training samples to obtain the three-dimensional clothing design drawing generation model, including: For any training sample among the plurality of training samples, obtaining a second text guide vector based on the second text information corresponding to the three-dimensional clothing design sample; Performing cross-attention processing on the starting word and the second text guide vector to obtain a third word sequence corresponding to the three-dimensional clothing design sample; Determining a fourth word sequence corresponding to the three-dimensional clothing design sample using a preset rule; A loss function is constructed based on the difference between the third word sequence and the fourth word sequence, and the initial three-dimensional clothing design drawing generation model is updated based on the loss function to obtain the three-dimensional clothing design drawing generation model.

7. The method according to claim 6, characterized in that The step of using a preset rule to determine the fourth word sequence corresponding to the three-dimensional clothing design sample includes: Determine the rotation quaternion and translation vector corresponding to each of the multiple polygonal panels corresponding to the three-dimensional clothing design sample, as well as the edge information and stitching information corresponding to the multiple edges of each polygonal panel; wherein the rotation quaternion is used to indicate the rotation angle of the polygonal panel relative to the global coordinate system; the translation vector is used to indicate the position offset of the polygonal panel relative to the global coordinate system; the edge information is used to indicate the starting point, end point and Bezier curve control point of the edge; the stitching information is used to indicate whether to stitch the edge and the position of the target edge to be stitched with the edge; The rotation quaternions and translation vectors corresponding to the multiple polygonal panels, and the edge information and stitching information corresponding to the multiple edges of each polygonal panel are spliced ​​to obtain the fourth word sequence.

8. A three-dimensional clothing design drawing generating device, characterized in that: The device comprises: an acquisition module, configured to acquire first text information, wherein the first text information is used to describe color information, style information, material information, and detail information of a desired first three-dimensional clothing design drawing; wherein the detail information is used to indicate at least one of a pattern, a collar shape, a sleeve shape, a skirt hem, and a trouser leg related to the first three-dimensional clothing design drawing; An output module is used to predict and process the first text information based on a three-dimensional clothing design drawing generation model to obtain and output the first three-dimensional clothing design drawing; wherein the three-dimensional clothing design drawing generation model is trained by three-dimensional clothing design drawing samples and second text information corresponding to the three-dimensional clothing design drawing samples, and the second text information is used to describe the color information, style information, material information and detail information of the three-dimensional clothing design drawing samples.

9. An electronic device, characterized in that: include: memory and at least one processor; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed, the method according to any one of claims 1 to 7 is implemented.

Images