Garment interactive design system and method based on artificial intelligence

Abstract

The invention relates to the field of costume design, and particularly discloses a costume interactive design system and method based on artificial intelligence, which uses an artificial intelligence technology based on the deep learning field to perform feature extraction and coding on a customer image, customer preference and a costume designer design sketch so as to generate a designed costume upper body effect picture. Therefore, the customer can directly observe whether the fabric, the design, the color matching and the like conform to the mind or not through intelligent generation of the designed garment upper body effect picture, understanding deviation caused by improper communication between the customer and a designer is avoided, and the satisfaction degree of the customer to customized garments is improved.

Description

Technical Field

[0001] The present application relates to the field of clothing design, and more specifically to an artificial intelligence-based clothing interactive design system and method. Background Art

[0002] Clothing is closely intertwined with people's lives. Clothing production methods primarily include traditional mass production and custom-made models. With the development of society and the general improvement of people's living standards, traditional mass production methods are no longer able to meet consumers' demand for fashion, variety, and personalization. More and more people are pursuing customization. The highly personalized customization model can meet customers' unique needs and showcase their individual beauty. Once custom clothing is produced, modifications can be difficult, especially those involving tailoring or structural adjustments. During the customization process, miscommunication between customers and tailors can lead to inconsistent understanding of clothing style, fabric, and fit. For example, customers lack understanding of the properties and effects of different fabrics, making it difficult to choose the right one.

[0003] Therefore, an optimized clothing design solution is expected. Summary of the Invention

[0004] This application is proposed to solve the above-mentioned technical problems. The embodiments of this application provide an artificial intelligence-based clothing interactive design system and method, which uses artificial intelligence technology based on deep learning to extract and encode customer images, customer preferences, and clothing designer design sketches to generate a rendering of the designed clothing. In this way, by intelligently generating the rendering of the designed clothing, customers can directly observe whether the fabric, design, color matching, etc. meet their expectations, avoiding misunderstandings caused by improper communication between customers and designers, and improving customer satisfaction with customized clothing.

[0005] According to one aspect of the present application, there is provided an artificial intelligence-based clothing interactive design system, which includes:

[0006] Clothing design data acquisition module, used to obtain customer images, customer preferences and clothing designer design sketches;

[0007] A design effect analysis module is used to analyze the customer preferences and the clothing designer's design sketches to obtain a global feature matrix of the design effect;

[0008] A customer image analysis module, configured to analyze the customer image and obtain a customer portrait feature vector;

[0009] The design clothing upper body effect generation module is used to comprehensively analyze the design effect global feature matrix and the customer portrait feature vector to generate the design clothing upper body effect diagram.

[0010] According to another aspect of the present application, there is also provided a clothing interactive design method based on artificial intelligence, which includes:

[0011] Obtain customer images, customer preferences, and clothing designer design sketches;

[0012] Analyzing the customer preferences and the clothing designer's design sketches to obtain a global feature matrix of design effects;

[0013] Analyze the customer image to obtain a customer portrait feature vector;

[0014] The global feature matrix of the design effect and the feature vector of the customer portrait are comprehensively analyzed to generate an effect diagram of the designed clothing on the body.

[0015] In summary, the AI-based interactive clothing design system and method provided by this application uses AI technologies based on deep learning to extract and encode features from customer images, customer preferences, and clothing designer sketches to generate renderings of the designed garments. This intelligent generation of renderings allows customers to directly assess whether the fabric, design, color scheme, and other aspects meet their needs, avoiding misunderstandings caused by improper communication between customers and designers and improving customer satisfaction with custom clothing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without inventive efforts. In the drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 4 is a block diagram of an artificial intelligence-based clothing interactive design system according to an embodiment of the present application.

[0018] Figure 2 4 is a block diagram of a design effect analysis module in an artificial intelligence-based clothing interactive design system according to an embodiment of the present application.

[0019] Figure 3 Flowchart of an artificial intelligence-based clothing interactive design method according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Below, in order to enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of this application.

[0021] Figure 1 FIG is a block diagram of an artificial intelligence-based clothing interactive design system according to an embodiment of the present application. Figure 1 As shown, the artificial intelligence-based clothing interactive design system 100 according to the embodiment of the present application includes: a clothing design data acquisition module 110, which is used to acquire customer images, customer preferences and clothing designer design sketches; a design effect analysis module 120, which is used to analyze the customer preferences and the clothing designer design sketches to obtain a global feature matrix of design effects; a customer image analysis module 130, which is used to analyze the customer image to obtain a customer portrait feature vector; and a designed clothing upper body effect generation module 140, which is used to comprehensively analyze the design effect global feature matrix and the customer portrait feature vector to generate a designed clothing upper body effect diagram.

[0022] In the aforementioned AI-based interactive clothing design system 100, the clothing design data acquisition module 110 is used to acquire customer images, customer preferences, and clothing designer design sketches. As discussed in the background art, once custom clothing is produced, modifications can be difficult, especially those involving cutting or structural adjustments. During the clothing customization process, there may be poor communication between the customer and the tailor, leading to inconsistent understanding of clothing style, fabric, and fit. For example, a customer lacks understanding of the properties and effects of different fabrics, making it difficult to select the appropriate fabric. Therefore, an optimized clothing design solution is desired.

[0023] To address these technical issues, we proposed an AI-based interactive clothing design system. This system uses deep learning-based AI techniques to extract and encode features from customer images, customer preferences, and designer sketches to generate images of how the designed garment would look. This intelligently generated image allows customers to directly assess whether the fabric, design, and color scheme meet their needs. This avoids misunderstandings that can arise from miscommunication between customers and designers, and improves customer satisfaction with custom clothing.

[0024] Currently, deep learning and neural networks are widely used in fields such as computer vision, natural language processing, and speech signal processing. Furthermore, deep learning and neural networks have demonstrated capabilities approaching or even surpassing those of humans in areas such as image classification, object detection, semantic segmentation, and text translation.

[0025] In recent years, the development of deep learning and neural networks has provided new solutions and solutions for clothing interactive design systems based on artificial intelligence.

[0026] Specifically, first, obtain customer images, customer preferences, and clothing designer design sketches. Customer images can help the system understand the customer's physical features and appearance, so as to customize clothing designs that suit the customer's body shape and style. The customer image should be slim-fitting, and the user obtains information such as the customer's size and skin color. Understanding the customer's preferences can help the system generate clothing designs that suit the customer's taste and style, thereby improving customer satisfaction. The customer's preferences include style, material, and color matching. The designer's sketches are a source of design inspiration and creativity, which can help the system better understand the designer's intentions and style, and ensure that the design effect is in line with the designer's original intention. By obtaining data such as customer images, customer preferences, and designer design sketches, the artificial intelligence-based clothing interactive design system can better understand customer needs and design intentions, thereby generating personalized clothing design effects that meet customer expectations.

[0027] In the above-mentioned artificial intelligence-based clothing interactive design system 100, the design effect analysis module 120 is used to analyze the customer preferences and the clothing designer's design sketches to obtain a global feature matrix of the design effect.

[0028] Figure 2 FIG is a block diagram of a design effect analysis module in an artificial intelligence-based clothing interactive design system according to an embodiment of the present application. Figure 2 As shown, the design effect analysis module 120 includes: a customer preference analysis unit 121, used to pass the customer preference through a converter-based customer preference semantic encoder to obtain multiple customer preference semantic feature vectors; a customer preference feature matrixing unit 122, used to arrange the multiple customer preference semantic feature vectors in two dimensions to obtain a customer preference feature matrix; a design sketch analysis unit 123, used to pass the clothing designer's design sketch through a design sketch feature extraction module based on a convolutional neural network model to obtain a design sketch feature matrix; and a design effect global fusion unit 124, used to fuse the customer preference feature matrix and the design sketch feature matrix to obtain a design effect global feature matrix.

[0029] Specifically, in an embodiment of the present application, the customer preference semantic encoder is a converter-based context encoder model including an embedding layer.

[0030] More specifically, customer preferences are converted into multiple semantic feature vectors using a transformer-based customer preference semantic encoder. This transformer-based context encoder model includes an embedding layer. Customer preferences are often multidimensional, encompassing aspects such as style, material, and color. Converting customer preferences into multiple semantic feature vectors allows for a more comprehensive capture of diverse customer preferences, improving understanding and modeling of customer needs. After converting customer preferences into semantic feature vectors, each feature vector represents a specific aspect of a customer's preference, enabling the system to more clearly understand diverse customer preferences and better factor these preferences into the subsequent design process. Furthermore, by decomposing customer preferences into multiple independent semantic feature vectors, confusion and interference between different preference factors can be avoided, ensuring that each feature vector independently influences the final design outcome, thereby improving the accuracy and personalization of the design results. The transformer-based context encoder model allows for flexible learning of semantic representations of customer preferences, adapting to the different ways and characteristics of preference expression among customers, thereby improving the model's generalization and adaptability, and making the system more universal and scalable.

[0031] Specifically, in an embodiment of the present application, the customer preference analysis unit 121 includes: an embedding conversion secondary sub-unit, which is used to pass the customer preferences through an embedding layer to convert the various dimensional data of the customer preferences into customer embedding vectors to obtain a sequence of customer embedding vectors, wherein the embedding layer uses a learnable embedding matrix to embed the various dimensional data of the customer preferences; an encoding secondary sub-unit, which is used to input the sequence of customer embedding vectors into the converter-based context encoder to obtain the multiple customer preference semantic feature vectors.

[0032] Specifically, in an embodiment of the present application, the encoding secondary sub-unit includes: a query vector construction tertiary sub-unit, which is used to arrange the sequence of customer embedding vectors into an input vector; a vector conversion tertiary sub-unit, which is used to convert the input vector into a query vector and a key vector respectively through a learnable embedding matrix; a self-attention tertiary sub-unit, which is used to calculate the product between the query vector and the transposed vector of the key vector to obtain a self-attention association matrix; a normalization tertiary sub-unit, which is used to normalize the self-attention association matrix to obtain a standardized self-attention association matrix; an attention calculation tertiary sub-unit, which is used to input the standardized self-attention association matrix into a Softmax activation function for activation to obtain a self-attention feature matrix; and an attention application tertiary sub-unit, which is used to multiply the self-attention feature matrix with each customer embedding vector in the sequence of customer embedding vectors to obtain the multiple customer preference semantic feature vectors.

[0033] More specifically, multiple customer preference semantic feature vectors are arranged in two dimensions to form a customer preference feature matrix. Arranging multiple customer preference semantic feature vectors into a matrix according to specific rules helps integrate the individual feature vectors into a cohesive structure, facilitating the system's holistic processing and analysis of customer preference information. The customer preference feature matrix provides a structured representation, allowing the system to more intuitively understand customer preference information. It also facilitates subsequent data processing and analysis, improving the readability and interpretability of the information.

[0034] More specifically, the fashion designer's design sketch is converted into a design sketch feature matrix using a design sketch feature extraction module based on a convolutional neural network model. Convolutional neural networks (CNNs) have excellent feature extraction capabilities in image processing. The CNN model can effectively extract visual features from design sketches. Integrating these visual features into a feature matrix helps better represent the key characteristics of the design sketch. Design sketches typically contain a large amount of pixel information, and the design sketch feature matrix can effectively reduce the dimensionality and compress this information, thereby reducing data complexity and redundancy, making subsequent processing and analysis more efficient.

[0035] Specifically, in an embodiment of the present application, the design sketch analysis unit 123 is used to: use each layer of the design sketch feature extraction module based on the convolutional neural network model to perform convolution processing, mean pooling processing along the channel dimension and nonlinear activation processing on the input data in the forward pass of the layer to output the design sketch feature matrix by the last layer of the design sketch feature extraction module based on the convolutional neural network model, wherein the input of the design sketch feature extraction module based on the convolutional neural network model is the design sketch of the clothing designer.

[0036] Furthermore, in order to combine customer needs and design features, the customer preference feature matrix and the design sketch feature matrix are fused to obtain a global feature matrix of design effects, thereby achieving a design effect that is more in line with customer preferences.

[0037] Specifically, in the embodiment of the present application, the design effect global fusion unit 124 is used to fuse the customer preference feature matrix and the design sketch feature matrix using the following cascade formula to obtain a design effect global feature matrix; wherein the cascade formula is:

[0038] M=Concat[M1,M2]

[0039] Among them, M1 represents the customer preference feature matrix, M2 represents the design sketch feature matrix, M represents the design effect global feature matrix, and Concat[·,·] represents the cascade function.

[0040] In the above-mentioned artificial intelligence-based clothing interactive design system 100, the customer image analysis module 130 is used to analyze the customer image to obtain a customer portrait feature vector.

[0041] Specifically, in the embodiment of the present application, the customer image analysis module 130 is used to: pass the customer portrait through a customer portrait feature extraction module based on a convolutional neural network model to obtain a customer portrait feature vector.

[0042] More specifically, customer profiles are converted into customer profile feature vectors using a customer profile feature extraction module based on a convolutional neural network model. Convolutional neural networks (CNNs) excel in image processing and can effectively extract features from images. Using the CNN model, visual features in customer profiles can be extracted and converted into more representative and abstract feature vectors. Furthermore, customer profiles may contain a large amount of pixel information, and converting them into feature vectors effectively reduces the data dimension and compresses the information volume, thereby reducing storage requirements and speeding up processing. Converting customer profiles into feature vectors makes the data easier to standardize and process. This helps unify the format of customer profile data, making subsequent analysis and application more convenient.

[0043] Specifically, in an embodiment of the present application, the customer image analysis module 130 is used to: use the various layers of the customer portrait feature extraction module based on the convolutional neural network model to perform the following on the input data in the forward pass of the layer: use the convolution units of the various layers of the customer portrait feature extraction module based on the convolutional neural network model to perform convolution processing on the input data to obtain a convolution feature map; use the pooling units of the various layers of the customer portrait feature extraction module based on the convolutional neural network model to perform pooling processing on the local feature matrix along the channel dimension of the convolution feature map to obtain a pooled feature map; and use the activation units of the various layers of the customer portrait feature extraction module based on the convolutional neural network model to perform nonlinear activation on the feature values ​​of each position in the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the customer portrait feature extraction module based on the convolutional neural network model is the customer portrait feature vector.

[0044] In the above-mentioned artificial intelligence-based clothing interactive design system 100, the designed clothing upper body effect generation module 140 is used to comprehensively analyze the design effect global feature matrix and the customer portrait feature vector to generate the designed clothing upper body effect diagram.

[0045] Specifically, in an embodiment of the present application, the designed clothing upper body effect generation module 140 includes: a clothing design effect upper body feature extraction unit, used to fuse the design effect feature matrix and the customer portrait feature vector to obtain a clothing design effect upper body feature vector; an optimization unit, used to perform multi-level internal structure optimization based on feature decomposition on the clothing design effect upper body feature vector to obtain an optimized clothing design effect upper body feature vector; and a designed clothing upper body effect generation unit, used to pass the optimized clothing design effect upper body feature vector through a generator to generate a designed clothing upper body effect diagram.

[0046] More specifically, to achieve personalized clothing design, the design effect feature matrix and the customer portrait feature vector are fused to obtain the clothing design effect upper body feature vector. The fusion of design effect features and customer portrait features enables personalized clothing design. By combining the customer's preferences and characteristics with the design effect, a clothing design solution that better meets the customer's needs and aesthetics is generated. Combining design effect features with customer portrait features can improve design accuracy, ensure that the design solution is consistent with customer expectations, and reduce the deviation between design and demand. At the same time, by fusing the customer portrait feature vector, the system can better understand the customer's preferences and characteristics. Combined with the design effect features, the clothing design effect upper body feature vector integrates information from design features and customer characteristics, providing a more comprehensive and integrated feature representation, which helps the system better understand the customer's needs and the overall requirements of the design task.

[0047] Specifically, in an embodiment of the present application, the clothing design effect upper body feature extraction unit is used to: fuse the design effect feature matrix and the customer portrait feature vector using the following fusion formula to obtain a clothing design effect upper body feature vector; wherein the fusion formula is:

[0048]

[0049] Among them, V1 represents the upper body feature vector of the clothing design effect, V2 represents the customer portrait feature vector, M c represents the design effect feature matrix, Represents matrix multiplication.

[0050] In particular, in the technical solution of the present application, on the one hand, considering that in the classification task of the upper body feature vector of clothing design effects, different types of clothing designs may have different sample numbers, for example, some clothing styles may appear more frequently in the data set, while other styles appear less frequently. The imbalance in the number of samples of different categories in the data set will cause the classification algorithm to tend to predict the category with a larger number of samples, because during the training process, the model is more likely to learn the feature representations of the category with a larger number of samples, while it will not learn enough for the category with a smaller number of samples. This bias will cause the classifier's prediction performance on the minority category to decline, thereby reducing the overall classification accuracy. Especially when faced with an unbalanced data set, it is easy to cause model bias and affect the classification effect. On the other hand, considering that the upper body feature vector of clothing design effects may contain a large number of features, some of these features may be redundant, that is, they do not provide additional information or are highly correlated with other features. Feature redundancy will increase the complexity of the model, reduce the generalization ability of the model, and may lead to overfitting problems. This will affect the performance of the classifier on new data, reducing its performance and generalization ability. Based on this, in the technical solution of the present application, the upper body feature vector of clothing design effects is optimized based on multi-level internal structure optimization of feature decomposition to obtain an optimized upper body feature vector of clothing design effects.

[0051] Specifically, the optimization unit is used to calculate the global multi-level autocorrelation structure matrix of the upper body feature vector of the clothing design effect, which is expressed as follows:

[0052]

[0053] M=D1⊙D2

[0054] v i ,v j ∈V

[0055] Among them, V represents the upper body feature vector of clothing design effect, v i and v j They represent the i-th and j-th eigenvalues ​​of the upper body feature vector of clothing design effects, w1, w2, w3 and w4 represent different weight hyperparameters, ⊙ represents matrix dot product, D1 represents the forward weighted matrix of the upper body feature of clothing design effects, and D2 represents the reverse weighted matrix of the upper body feature of clothing design effects. Represents the value of the (i, j)th position of the positive weighted matrix of the upper body feature of the clothing design effect, It represents the value of the (i, j)th position of the inverse weighted matrix of the upper body features of the clothing design effect, and M represents the global multi-level autocorrelation structure matrix.

[0056] That is, by explicitly capturing and quantifying the statistical dependencies or linear correlations of the components within the upper body feature vector of clothing design effects at different levels, the multi-level structural information of all dimensions or elements of the upper body feature vector of clothing design effects is covered, and the originally implicit multi-level structural information is converted into a mathematical object that can be directly analyzed, laying the foundation for subsequent operations such as feature decomposition.

[0057] Specifically, the optimization unit is further used to perform eigendecomposition on the global multi-level autocorrelation structure matrix to obtain a set of multi-level feature component coding vectors of the clothing design effect upper body, which is expressed as follows:

[0058]

[0059] Where Λ represents a diagonal matrix, λ1 and λ m denote the first and mth eigenvalues ​​of the diagonal matrix, respectively, (·) T represents the transpose of the vector, U represents the set of multi-level feature component encoding vectors of the upper body of clothing design effects, x1, x2 and x m They respectively represent the first, second and mth clothing design effect upper body multi-level feature component encoding vectors in the set of clothing design effect upper body multi-level feature component encoding vectors.

[0060] That is, through feature decomposition, the complex correlation structure within the global multi-level autocorrelation structure matrix is ​​effectively decoupled, and a set of multi-level feature component encoding vectors of the clothing design effect in the form of an orthogonal basis guided by the internal structure of the data is obtained. It represents different independent patterns of internal changes in the features and is sorted by importance, providing a basis for subsequent targeted adjustments to each independent structural aspect, thereby more accurately integrating the design effect and customer portrait characteristics.

[0061] Specifically, the optimization unit is further used to: perform data compression on each clothing design effect upper body multi-level feature component coding vector in the set of clothing design effect upper body multi-level feature component coding vectors to obtain a set of clothing design effect upper body multi-level feature component adjustment coding vectors, which is expressed as follows:

[0062]

[0063] Among them, x i represents the i-th clothing design effect upper body multi-level feature component encoding vector in the set of clothing design effect upper body multi-level feature component encoding vectors, ||·|| represents the Euclidean norm, y i Represents the regulated encoding vector of the multi-level feature component of the upper body of the i-th clothing design effect.

[0064] That is, data compression is performed on the encoding vectors of the multi-level feature components of the clothing design effect on the body. This is based on the decomposition of independent structural patterns, and further through transformation operations such as adjusting the vector norm and capturing high-order statistical dependencies, each independent multi-level feature component of the clothing design effect on the body is refined. On the one hand, redundant information is compressed and core structural features are retained, so that the encoding vectors of the multi-level feature components of the clothing design effect on the body are more compact and focus on key information. On the other hand, nonlinear transformation is used to enhance the discrimination ability of the multi-level feature components of the clothing design effect on the body, suppress noise interference, and break through the limitations of pure linear decomposition.

[0065] Specifically, the optimization unit is further used to calculate the internal significance adjustment factor of each clothing design effect upper body multi-level feature component adjustment coding vector in the set of clothing design effect upper body multi-level feature component adjustment coding vectors to obtain a set of internal significance adjustment factors, which is expressed by the formula:

[0066]

[0067] Among them, α and β represent different weight parameters, ||·||1 represents the first norm, ||·||2 represents the second norm, L represents the length of the multi-level feature component adjustment coding vector of the clothing design effect, a i represents y i The corresponding internal significance moderator.

[0068] That is, by dynamically measuring the contribution of the multi-level feature components of the upper body of each clothing design effect to the overall design effect, the importance weight basis based on the real-time state is provided for subsequent fusion, so that the model can adaptively focus on key features and weaken secondary information, ensuring that when the design effect and customer portrait features are integrated, the generated upper body effect diagram can better highlight the matching degree between the core design elements and the customer's body shape, avoid the visual presentation deviation caused by the ambiguity of feature importance, and improve the accuracy of customized design and the customer's intuitive understanding of the design effect from the feature weighting level, effectively reducing information loss and cognitive differences in communication.

[0069] Specifically, the optimization unit is further configured to: normalize the set of internal significance adjustment factors to obtain a set of internal significance adjustment weights, which is expressed as:

[0070] w i =Softmax(a i )

[0071] Among them, Softmax(·) represents the normalization function, w i Indicates a i The corresponding internal significance adjustment weight.

[0072] Specifically, through regularization, namely the softmax function, the set of internal significance adjustment factors is converted into a set of internal significance adjustment weights in the form of a probability distribution summing to 1. This unifies the scale and controls the sharpness of the weight distribution, avoiding instability in the fusion process caused by differences in numerical ranges or extreme values, and ensuring the numerical stability and controllability of the model during feature fusion. Furthermore, through normalization, each feature component is assigned a dynamic and balanced importance weight, enabling subsequent fusion to adaptively integrate different feature components based on the normalized weights, precisely focusing on high-value structural information and reducing interference from less important features.

[0073] Specifically, the optimization unit is further configured to: based on the set of internal significance adjustment weights, perform fine-grained fusion on the set of multi-level feature component adjustment coding vectors of the clothing design effect upper body to obtain an optimized clothing design effect upper body feature vector, which is expressed as follows:

[0074]

[0075] Among them, V′ represents the upper body feature vector of the optimized clothing design effect.

[0076] That is, through fine-grained fusion, the multi-level feature component adjustment coding vectors of the clothing design effect upper body can be weighted and combined based on the internal saliency adjustment weights, highlighting key information and suppressing secondary or noisy information. Through this fusion method, the representational ability of the multi-level feature component adjustment coding vectors of the clothing design effect upper body is enhanced. The resulting optimized clothing design effect upper body feature vector condenses the key structural information within the clothing design effect upper body features, making it more robust and more discriminative than the clothing design effect upper body feature vector.

[0077] The optimized upper-body feature vectors of the garment design are then used to generate upper-body renderings of the garment. Generating upper-body renderings through the generator allows for a visual display of the design, allowing designers and clients to more intuitively understand the appearance and coordination of garments, thereby better evaluating the suitability of design proposals. Generating upper-body renderings of the garment can be used to present design proposals to clients, providing a clearer understanding of the design concept, improving communication efficiency, and reducing misunderstandings and unnecessary modifications. This can also reduce the cost and time of producing actual garment samples, allowing for a more proactive demonstration of design results and avoiding unnecessary cost and time.

[0078] Specifically, in an embodiment of the present application, the generator uses a generative adversarial network (GAN), which is composed of a generator and a discriminator and can be used to generate a design clothing upper body effect image. In the technical solution of the present application, the generator can receive the optimized clothing design effect upper body feature vector as input and generate the design clothing upper body effect image.

[0079] In summary, the AI-based interactive clothing design system according to the embodiments of this application has been explained. It uses AI technology based on deep learning to extract and encode features from customer images, customer preferences, and clothing designer design sketches to generate renderings of the designed garments. This intelligent generation of renderings allows customers to directly observe whether the fabric, design, color scheme, and other aspects meet their needs, avoiding misunderstandings caused by improper communication between customers and designers and improving customer satisfaction with customized clothing.

[0080] As described above, the artificial intelligence-based clothing interactive design system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as an artificial intelligence-based clothing interactive design server. In one example, the artificial intelligence-based clothing interactive design system 100 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or hardware module. For example, the artificial intelligence-based clothing interactive design system 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the artificial intelligence-based clothing interactive design system 100 can also be one of the many hardware modules of the terminal device.

[0081] Alternatively, in another example, the artificial intelligence-based clothing interaction design system 100 and the terminal device may also be separate devices, and the artificial intelligence-based clothing interaction design system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interaction information in accordance with an agreed data format.

[0082] Based on the same inventive concept, an embodiment of the present application also provides an artificial intelligence-based clothing interaction design method, which can be used to implement the system described in the above embodiment, as described in the following embodiment.

[0083] Figure 3 Flowchart of the clothing interactive design method based on artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the artificial intelligence-based clothing interactive design method according to the embodiment of the present application includes the steps of: S110, acquiring customer images, customer preferences and clothing designer design sketches; S120, analyzing the customer preferences and the clothing designer design sketches to obtain a global feature matrix of design effects; S130, analyzing the customer images to obtain a customer portrait feature vector; and S140, comprehensively analyzing the design effect global feature matrix and the customer portrait feature vector to generate a design clothing upper body effect diagram.

[0084] Here, those skilled in the art will understand that the specific operations of each step in the above-mentioned clothing interactive design method based on artificial intelligence have been referred to above. Figures 1 to 2 The description of the artificial intelligence-based clothing interactive design system has been introduced in detail, and therefore, its repeated description will be omitted.

[0085] In summary, the AI-based interactive clothing design method according to the embodiments of this application has been explained. It uses AI technology based on deep learning to extract and encode features from customer images, customer preferences, and clothing designer design sketches to generate renderings of the designed garment. This intelligent generation of renderings allows customers to directly observe whether the fabric, design, color scheme, and other aspects meet their needs, avoiding misunderstandings caused by improper communication between customers and designers and improving customer satisfaction with customized clothing.

[0086] References to "first," "second," "third," "fourth," and the like (if any) herein are intended to distinguish similar objects and are not necessarily intended to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions, e.g., a process, method, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, or apparatus.

[0087] It should be noted that the descriptions of "first", "second", etc. in this application are for descriptive purposes only and should not be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0088] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

[0089] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0090] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0091] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0092] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. An artificial intelligence-based clothing interactive design system, characterized in that: include: Clothing design data acquisition module, used to obtain customer images, customer preferences and clothing designer design sketches; A design effect analysis module is used to analyze the customer preferences and the clothing designer's design sketches to obtain a global feature matrix of the design effect; A customer image analysis module, configured to analyze the customer image and obtain a customer portrait feature vector; The design clothing upper body effect generation module is used to comprehensively analyze the design effect global feature matrix and the customer portrait feature vector to generate the design clothing upper body effect diagram.

2. The artificial intelligence-based clothing interactive design system according to claim 1, characterized in that: The design effect analysis module includes: A customer preference analysis unit, configured to pass the customer preferences through a converter-based customer preference semantic encoder to obtain a plurality of customer preference semantic feature vectors; a customer preference feature matrixing unit, configured to perform two-dimensional arrangement on the plurality of customer preference semantic feature vectors to obtain a customer preference feature matrix; A design sketch analysis unit, configured to pass the clothing designer's design sketch through a design sketch feature extraction module based on a convolutional neural network model to obtain a design sketch feature matrix; The design effect global fusion unit is used to fuse the customer preference feature matrix and the design sketch feature matrix to obtain a design effect global feature matrix.

3. The artificial intelligence-based clothing interactive design system according to claim 2, characterized in that: The customer image analysis module is used to: The customer portrait is passed through a customer portrait feature extraction module based on a convolutional neural network model to obtain a customer portrait feature vector.

4. The artificial intelligence-based clothing interactive design system according to claim 3, characterized in that: The design clothing upper body effect generation module includes: A clothing design effect upper body feature extraction unit, configured to fuse the design effect feature matrix and the customer portrait feature vector to obtain a clothing design effect upper body feature vector; an optimization unit, configured to perform multi-level internal structure optimization based on feature decomposition on the clothing design effect upper body feature vector to obtain an optimized clothing design effect upper body feature vector; The design clothing upper body effect generation unit is used to pass the optimized clothing design effect upper body feature vector through a generator to generate a design clothing upper body effect image.

5. The artificial intelligence-based clothing interactive design system according to claim 4, characterized in that: The optimization unit is used to: Calculating a global multi-level autocorrelation structure matrix of the upper body feature vector of the clothing design effect; Performing eigendecomposition on the global multi-level autocorrelation structure matrix to obtain a set of multi-level feature component coding vectors of the clothing design effect upper body; performing data compression on each clothing design effect upper body multi-level feature component coding vector in the set of clothing design effect upper body multi-level feature component coding vectors to obtain a set of clothing design effect upper body multi-level feature component adjustment coding vectors; Calculating the internal significance adjustment factor of each clothing design effect upper body multi-level feature component adjustment coding vector in the set of clothing design effect upper body multi-level feature component adjustment coding vectors to obtain a set of internal significance adjustment factors; Normalizing the set of internal significance adjustment factors to obtain a set of internal significance adjustment weights; Based on the set of internal significance adjustment weights, the set of clothing design effect upper body multi-level feature component adjustment coding vectors is fine-grainedly fused to obtain an optimized clothing design effect upper body feature vector.

6. A clothing interactive design method based on artificial intelligence, characterized in that: include: Obtain customer images, customer preferences, and clothing designer design sketches; Analyzing the customer preferences and the clothing designer's design sketches to obtain a global feature matrix of design effects; Analyze the customer image to obtain a customer portrait feature vector; The global feature matrix of the design effect and the feature vector of the customer portrait are comprehensively analyzed to generate an effect diagram of the designed clothing on the body.

Images