Method for predicting properties of a knitted product and design method

Abstract

The application provides a performance prediction method and device of a knitted product, a design method of the knitted product, and a corresponding computer readable storage medium. The performance prediction method comprises the following steps: obtaining overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product, wherein the overall style data in the multiple dimensions at least comprises visual data of the knitted product; dividing the knitted product into multiple regions according to the knitting structure data, wherein each region adopts one kind of knitting structure; determining structure information representation of each region according to the knitting structure data and the yarn data; determining style information representation of the knitted product according to the overall style data in the multiple dimensions; and predicting at least one performance parameter of the knitted product according to the structure information representation and the style information representation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knitting, and in particular to a performance prediction method of a knitted product, a design method of a knitted product, a performance prediction device of a knitted product, a design device of a knitted product, and a corresponding computer-readable storage medium. BACKGROUND

[0002] In the design process of a knitted product, a designer cannot accurately predict or calculate the product performance, and can only rely on personal experience or intuition to design the product, and then obtains accurate product performance parameters through trial production and actual testing. Generally, when a designer designs a new style of knitted product, there is a large difference between the actual performance parameters of the obtained sample and the target performance parameters assumed by the designer when designing the product. This requires the designer to adjust various parameters such as yarn material specifications, fabric organization, and knitting process, and continuously make samples and test until the performance parameters of the sample meet expectations.

[0003] In addition, in order to represent the attributes of a garment or its accessories, there is an existing technology in the technical field of online shopping and the like, which extracts visual features of a garment or its accessories by recognizing a product image, and then predicts the attributes of the garment or its accessories through a pre-trained deep neural network model. However, this prediction method can only make amateur-level prediction of attributes such as style, shape, texture, color, and fabric characteristics based on visual features such as color, texture, and contour shape in the product image, thereby providing a preliminary reference for consumers to select and purchase products, but cannot meet the needs of designers for multi-dimensional performance parameter prediction and professional-level prediction accuracy of knitted products.

[0004] In order to overcome the above-mentioned defects of the prior art, there is an urgent need in the art for a performance prediction technology of a knitted product for realizing multi-dimensional prediction of performance parameters of a knitted product and improving the prediction accuracy of performance parameters, so as to reduce the workload of trial production, sample making, performance parameter testing and the like in the design process of a knitted product, thereby shortening the product design cycle, improving the product design efficiency, and improving the environmental efficiency of product design. SUMMARY

[0005] The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later.

[0006] In order to overcome the above-mentioned defects existing in the prior art, the present application provides a knitted product performance prediction method, a knitted product design method, a knitted product performance prediction device, a knitted product design device, and a corresponding computer readable storage medium, which can realize multi-dimensional prediction of performance parameters of a knitted product, improve the prediction accuracy of performance parameters, reduce the workload of trial production, performance parameter testing and other links in the design process of the knitted product, thereby shortening the product design cycle, improving the product design efficiency, and improving the environmental efficiency of product design.

[0007] Specifically, the above-mentioned knitted product performance prediction method according to the first aspect of the present application comprises the following steps: obtaining overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product, wherein the overall style data in multiple dimensions at least includes visual data of the knitted product; dividing the knitted product into multiple regions according to the knitting structure data, wherein each region adopts a knitting structure; determining structure information representation of each region according to the knitting structure data and the yarn data; determining style information representation of the knitted product according to the overall style data in multiple dimensions; and predicting at least one performance parameter of the knitted product according to the structure information representation and the style information representation.

[0008] Further, in some embodiments of the present application, the step of determining the structure information representation of each region according to the knitting structure data and the yarn data comprises: disassembling the region into multiple pixel points according to positions; representing the knitting structure type of each pixel point by a first multi-dimensional vector s ij ; representing the yarn data of each pixel point by a second multi-dimensional vector y ij ; determining the structure information vector t ij of each pixel point according to the first multi-dimensional vector s ij and the second multi-dimensional vector y ij ; and combining the structure information vectors t ij of each pixel point according to positions to determine the structure information representation T of the knitted product in the region.

[0009] Further, in some embodiments of the present application, the step of representing the knitting structure type of each pixel point by a first multi-dimensional vector s ij comprises representing the knitting structure type of each pixel point by a first multi-dimensional one-hot vector s ij .

[0010] Further, in some embodiments of the present application, the yarn data includes yarn material and yarn count. The second multi-dimensional vector y ijThe step of representing the yarn data of each pixel point includes: representing the yarn material of each pixel point by a second multi-dimensional one-hot vector; representing the yarn count of each pixel point by second normalized data; and splicing the second multi-dimensional one-hot vector and the second normalized data to determine the second multi-dimensional vector y representing the yarn data of each pixel point ij .

[0011] Further, in some embodiments of the present application, the step of combining the structure information vectors t of each pixel point according to the position ij to determine the structure information representation T of the knitting product in the region includes: combining the structure information vectors t of each pixel point according to the position ij to determine the structure initialization representation T0 of each region of the knitting product; and inputting the structure initialization representation T0 into a pre-trained structure information representation, performing at least one convolution operation on the structure initialization representation T0 according to the first learning parameters therein, to determine the structure information representation T including at least one-dimensional latent feature.

[0012] Further, in some embodiments of the present application, the visual data includes at least one of style data, size data, shape data of the knitting product. In addition, the overall style data of the plurality of dimensions further includes yarn data of at least one yarn involved in the knitting product, and / or device data of at least one knitting device.

[0013] Further, in some embodiments of the present application, the step of determining the style information representation D of the knitting product according to the overall style data of the plurality of dimensions includes: representing the style of the knitting product by a third multi-dimensional one-hot vector, and / or representing the size of the knitting product by fourth normalized data, and / or representing the shape of the knitting product by a fifth multi-dimensional one-hot vector, and / or representing at least one yarn material and / or yarn attribute involved in the knitting product by a sixth multi-dimensional vector / matrix, and / or representing at least one knitting device type and / or needle cylinder size involved in the knitting product by a seventh multi-dimensional vector / matrix; and splicing the third multi-dimensional one-hot vector, the fourth normalized data, the fifth multi-dimensional one-hot vector, the sixth multi-dimensional vector / matrix, and / or the seventh multi-dimensional vector / matrix to determine the style information representation D of the knitting product.

[0014] Further, in some embodiments of the present application, the step of concatenating the third multi-dimensional one-hot vector, the fourth normalized data, the fifth multi-dimensional one-hot vector, the sixth multi-dimensional vector and / or the seventh multi-dimensional vector to determine the style information representation D of the knitted product comprises: concatenating the third multi-dimensional one-hot vector, the fourth normalized data, the fifth multi-dimensional one-hot vector, the sixth multi-dimensional vector and / or the seventh multi-dimensional vector to determine a style initialization representation D0 of the knitted product; and inputting the style initialization representation D0 into a pre-trained style information determinator to perform at least one convolution operation on the style initialization representation D0 according to second learning parameters therein to determine the style information representation D comprising at least one-dimensional latent feature.

[0015] Further, in some embodiments of the present application, the step of predicting at least one performance parameter of the knitted product according to the structure information representation and the style information representation comprises: concatenating the structure information representation T and the style information representation D to determine a performance prediction initialization representation P0 of the knitted product; inputting the performance prediction initialization representation P0 into a pre-trained at least one performance parameter predictor to perform at least one convolution operation on the performance prediction initialization representation P0 according to corresponding third learning parameters to determine a performance prediction representation P comprising at least one-dimensional latent feature; and determining at least one performance parameter corresponding to the performance parameter predictor according to the performance prediction representation P.

[0016] Further, in some embodiments of the present application, the at least one performance parameter comprises at least one of a product size, a shrinkage rate, a weight, an elastic recovery rate, a moisture absorption and perspiration coefficient, a softness, a shear coefficient, and a deformation coefficient of the knitted product.

[0017] In addition, the above-mentioned design method of the knitted product according to the second aspect of the present application comprises the following steps: obtaining design data of a knitted product, wherein the design data comprises overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product; implementing the above-mentioned performance prediction method of the knitted product of the first aspect of the present application to determine at least one performance parameter of the knitted product; judging whether the performance of the knitted product meets the expectation according to the at least one performance parameter; and in response to the judgment result that the performance of the knitted product does not meet the expectation, modifying the design data of the knitted product.

[0018] Further, in some embodiments of the present application, the design method further comprises the following steps: in response to a result of judging that the performance of the knitted product meets the expectation, manufacturing a sample of the knitted product according to the design data; testing the sample to obtain measured data of the sample; verifying whether the performance of the knitted product meets the expectation according to the measured data; in response to a result of verifying that the performance of the knitted product does not meet the expectation, recording the design data of the knitted product and the measured data of the sample as a data sample of a performance predictor of the knitted product, and / or modifying the design data of the knitted product.

[0019] Further, the performance prediction device of the knitted product according to the third aspect of the present application comprises a memory and a processor. The processor is connected to the memory and is configured to implement the performance prediction method of the knitted product according to the first aspect of the present application.

[0020] Further, the design device of the knitted product according to the fourth aspect of the present application comprises a memory and a processor. The processor is connected to the memory and is configured to implement the design method of the knitted product according to the second aspect of the present application.

[0021] Further, the computer readable storage medium according to the fifth aspect of the present application has computer instructions stored thereon. The computer instructions are executed by a processor to implement the performance prediction method of the knitted product according to the first aspect of the present application.

[0022] Further, the computer readable storage medium according to the sixth aspect of the present application has computer instructions stored thereon. The computer instructions are executed by a processor to implement the design method of the knitted product according to the second aspect of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0023] The above features and advantages of the present application can be better understood by reading the detailed description of embodiments of the present application in conjunction with the following drawings, in which: in the drawings, components are not necessarily drawn to scale, and components having similar related functions or features can have the same or similar reference label.

[0024] Figure 1 A flowchart of a design method of a knitted product according to some embodiments of the present application is shown.

[0025] Figure 2 A flowchart of a performance prediction method of a knitted product according to some embodiments of the present application is shown.

[0026] Figure 3 A flowchart of a performance prediction method of a knitted product according to some embodiments of the present application is shown. Detailed Implementation

[0027] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0028] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0029] As mentioned above, in the design process of knitted products, designers cannot accurately predict or calculate product performance. They can only rely on personal experience or intuition to design products and then obtain accurate product performance parameters through sample production and actual testing. Generally, when designers design new styles of knitted products, the actual performance parameters of the samples they obtain often differ significantly from the target performance parameters they envisioned during the design process. This requires designers to adjust various parameters such as yarn specifications, fabric structure, and knitting techniques multiple times, and to continuously conduct sampling and testing until the performance parameters of the samples meet expectations. Furthermore, existing prediction methods can only make amateur-level predictions of attributes such as style, shape, texture, color, and fabric characteristics based on visual features such as color, texture, and outline shape in product images. This provides consumers with a preliminary reference for purchasing products, but it cannot meet designers' needs for multi-dimensional performance parameter prediction and professional-level prediction accuracy for knitted products.

[0030] In order to overcome the above-mentioned defects in the prior art, the present application provides a knitted product performance prediction method, a knitted product design method, a knitted product performance prediction device, a knitted product design device, and a corresponding computer readable storage medium, which can realize multi-dimensional prediction of performance parameters of the knitted product, improve the prediction accuracy of the performance parameters, reduce the workload of trial production, performance parameter testing and other links in the design process of the knitted product, thereby shortening the product design cycle, improving the product design efficiency, and improving the environmental efficiency of product design.

[0031] In some non-limiting embodiments, the above-mentioned knitted product performance prediction method provided by the first aspect of the present application can be implemented via the above-mentioned knitted product performance prediction device provided by the third aspect of the present application. Specifically, the performance prediction device is configured with a first memory and a first processor. The first memory includes but is not limited to the above-mentioned computer readable storage medium provided by the fifth aspect of the present application, on which computer instructions are stored. The first processor is connected to the first memory and is configured to execute the computer instructions stored on the first memory to implement the above-mentioned knitted product performance prediction method provided by the first aspect of the present application.

[0032] In addition, the above-mentioned knitted product design method provided by the second aspect of the present application can be implemented via the above-mentioned knitted product design device provided by the fourth aspect of the present application. Specifically, the design device is configured with a second memory and a second processor. The second memory includes but is not limited to the above-mentioned computer readable storage medium provided by the sixth aspect of the present application, on which computer instructions are stored. The second processor is connected to the second memory and is configured to execute the computer instructions stored on the second memory to implement the above-mentioned knitted product design method provided by the second aspect of the present application.

[0033] Please refer to Figures 1-3 . Figure 1 A flowchart of a knitted product design method according to some embodiments of the present application is shown. Figure 2 A flowchart of a knitted product performance prediction method according to some embodiments of the present application is shown. Figure 3 A flowchart of a knitted product performance prediction method according to some embodiments of the present application is shown.

[0034] As Figure 1As shown, in the design process of seamless knitted products produced via seamless machines or hosiery machines, a designer can first complete a design file (e.g., a DIS BMP document) of the knitted product by using a design program. The DIS BMP picture design document not only contains a style picture of the knitted product, but also defines various structural information of the knitted product at each position, such as the knitting structure type, the yarn material, the yarn count, and the like. In response to the designer completing and exporting the design file of the knitted product, the design device can obtain the design data of the knitted product from the design file and transmit the design data to the performance prediction device provided in the third aspect of the present application to predict at least one performance parameter of the knitted product.

[0035] As shown, Figures 1-3 After obtaining the design data of the knitted product, the performance prediction device can analyze the design data to obtain overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product.

[0036] In some embodiments, the overall style data includes but is not limited to visual data that can be directly identified and obtained from the style picture, such as the style type, the size, the contour shape, and the like of the knitted product, yarn data of at least one yarn involved in the knitted product, such as the yarn material, the count, and the like, and / or device data of at least one knitting device required for producing the knitted product, such as the device type, the needle cylinder size, and the like. The performance prediction device can represent the type data in a One-Hot manner, and process the numerical information in a normalized manner to represent it as a decimal number in the interval (0, 1).

[0037] For example, the performance prediction device can represent the style of the knitted product by using a multi-dimensional One-Hot vector st, where each dimension of the vector st corresponds to a selectable style type. For another example, the performance prediction device can represent the size of the knitted product by using normalized data sc, where the data sc is a decimal number in the interval (0, 1). For another example, the performance prediction device can represent the shape of the knitted product by using a multi-dimensional One-Hot vector sh, where each dimension of the vector sh corresponds to a selectable shape. For another example, the performance prediction device can represent at least one yarn material and / or yarn count and the like attributes of the knitted product by using a multi-dimensional vector y or a matrix Y, where a part of the dimensions of the vector y or the matrix Y correspond to selectable yarn materials, and the other part of the dimensions are decimal numbers in the interval (0, 1), indicating the count of the corresponding yarn material. For another example, the performance prediction device can represent the type and / or the needle cylinder size and the like attributes of at least one knitting device required for producing the knitted product by using a multi-dimensional vector e or a matrix E, where a part of the dimensions of the vector e or the matrix E correspond to selectable device types, and the other part of the dimensions are decimal numbers in the interval (0, 1), indicating the needle cylinder size of the corresponding device type.

[0038] Afterwards, the performance prediction device can concatenate the overall style data of the knit product in each dimension, i.e., one or more of the above style vector st, size data sc, shape vector sh, yarn vector y / yarn matrix Y, and equipment vector e / equipment matrix E, to determine the style information representation D of the knit product.

[0039] Specifically, as shown in Figure 3 the process of determining the style information representation D, the performance prediction device can first concatenate one or more of the above style vector st, size data sc, shape vector sh, yarn vector y / yarn matrix Y, and equipment vector e / equipment matrix E to determine the style initialization representation Do of the knit product. Afterwards, the performance prediction device can input the style initialization representation Do into the input layer of the pre-trained style information representation, and pass the data via the input layer to one or more hidden layers of the backend. In some embodiments, the one or more hidden layers can be selected from a convolutional neural network. In response to the multi-dimensional data transmitted by the previous layer of neural network, each convolutional neural network will perform a convolution operation on several data in the style initialization representation Do according to the second learning parameters determined by pre-training, so as to obtain the style information representation D including at least one-dimensional latent feature at the output layer of the style information representation after at least one convolution operation.

[0040] In addition, as shown in Figure 1 and Figure 2 , for the knit structure data and yarn data of multiple positions of the knit product, the performance prediction device can divide the knit product into multiple regions each adopting one type of knit structure according to the knit structure data of each position, and represent the knit structure of each region respectively.

[0041] Specifically, in the process of determining the structure information representation of each region, the performance prediction device can first disassemble each region into multiple pixels according to the position based on the style picture in the DIS BMP picture design document. Afterwards, for each pixel, the performance prediction device can represent the knit structure type of the pixel by a multi-dimensional vector s ij , and represent the yarn data of the pixel by a multi-dimensional vector y ij .

[0042] In some embodiments, the multi-dimensional vector s ij may be in the form of a multi-dimensional One-Hot vector, where each dimension corresponds to one type of selectable knit structure. For example, for a seamless knit product woven by three types of knit stitches, the selectable knit structure type is three, and the knit stitch used by each pixel can only be one of the three types. Thus, the vector s ijcorrespondingly set as a three-dimensional one-hot vector (for example: (1, 0, 0)) to represent the knitting structure type of each pixel point. The performance prediction device can initialize and encode the knitting structure data of each pixel point to obtain a corresponding knitting structure matrix In the formula, n*m represents a sufficiently large rectangular region composed of n horizontal pixel points and m vertical pixel points, which can cover the entire knitting product.

[0043] In addition, in some embodiments, the multi-dimensional vector y ij The multi-dimensional one-hot vector can be used to splice the normalized data, the first or last dimension of which can be a decimal in the interval (0, 1) to indicate the yarn count, and the remaining dimensions can correspond to one of the optional yarn materials respectively. For example, for a knitting area made of four kinds of yarn materials, the optional yarn material types are four, and the yarn material used by each pixel point can only be one of the four materials. In this way, the vector y ij correspondingly set as a five-dimensional one-hot vector (for example: (1, 0, 0, 0, 0.5)) to represent the yarn material type and the yarn count of each pixel point respectively. The performance prediction device can initialize and encode the yarn data of each pixel point to obtain a corresponding yarn information matrix In the formula, n*m represents a sufficiently large rectangular region composed of n horizontal pixel points and m vertical pixel points, which can cover the entire knitting product.

[0044] Then, the performance prediction device can determine the structure information vector t ij of each pixel point according to the first multi-dimensional vector s ij and the second multi-dimensional vector y ij , and combine the structure information vectors t ij of each pixel point according to their positions to determine the structure information representation T of the knitting product in each region.

[0045] Specifically, in the process of determining the structure information representation T, the performance prediction device can first fill the knitting stitch one-hot vector s ij corresponding to each pixel point, and fill the knitting stitch one-hot vector s ij to the same dimension as the yarn material representation vector. For example, assuming that y ij is an l-dimensional vector, the value of l is determined by the number of yarn materials and the number of yarn counts considered, then the structure information vector t ij of a pixel point can be represented as a 2*l-dimensional matrix. In this way, for the entire knitting product, the performance prediction device can combine the initialization representations t ijThe structure initialization tensor T0 is formed by splicing the positions to represent the structure information of the entire knitted product in an n*m*2*l dimension. In the formula, n*m represents a large enough rectangular area composed of n horizontal pixel points and m vertical pixel points, which can cover the entire knitted product.

[0046] Then, as shown in Figure 3 , the performance prediction device can input the structure initialization representation T0 into the input layer of the pre-trained structure information representor, and transmit data to one or more hidden layers of the back end via the input layer. In some embodiments, the one or more hidden layers can be selected from a convolutional neural network. In response to the multi-dimensional data transmitted by the previous layer of neural network, each convolutional neural network will perform convolution operation on the structure information of several adjacent pixel points in the structure initialization representation T0 according to the first learning parameters determined by pre-training, so as to obtain the structure information representation T including at least one-dimensional latent feature in the output layer of the structure information representor after at least one convolution operation. By combining the DIS BMP picture design document to pixel-level disassemble and represent the knitted product, the present application can effectively extract the knitted structure information of the knitted product and the potential information between each knitted structure, so as to further improve the accuracy of performance prediction.

[0047] Then, the performance prediction device can predict at least one performance parameter of the knitted product according to the structure information representation T and the style information representation D. Here, the at least one performance parameter includes but is not limited to at least one of the product size, the shrinkage rate, the weight, the elastic recovery rate, the moisture absorption and perspiration coefficient, the softness, the shear coefficient and the deformation coefficient of the knitted product. Correspondingly, as shown in Figure 2 , the performance prediction device can be preferably configured with at least one performance parameter predictor for predicting the corresponding performance parameter.

[0048] Specifically, in the process of predicting the performance parameter of the knitted product, the performance prediction device can first splice the structure information representation T and the style information representation D to determine the performance prediction initialization representation P0 of the knitted product, and then as shown in Figure 2 and Figure 3The performance prediction initialization representation P0 is input into the input layer of the pre-trained performance parameter predictor, and data is transmitted to one or more hidden layers of the back end via the input layer. In some embodiments, the one or more hidden layers can be selected from a convolutional neural network. In response to the multi-dimensional data transmitted by the previous layer of the neural network, each convolutional neural network will perform convolution operations on the data of several adjacent pixel points in the prediction initialization representation P0 according to third learning parameters determined in advance, so that after at least one convolution operation, a structure information representation P including at least one-dimensional latent features is obtained at the output layer of the performance parameter predictor. Then, the performance prediction device can input the performance prediction representation P into the pre-trained classification network, and determine the predicted value label of the corresponding performance parameter according to fourth learning parameters determined in advance.

[0049] Similarly, the performance prediction device can also input the performance prediction initialization representation P0 of the knitted product into the predictor of each performance parameter such as product size, shrinkage rate, weight, elastic recovery rate, moisture absorption and perspiration coefficient, softness, shear coefficient, deformation coefficient, etc., to accurately predict various performance parameters, which will not be repeated here.

[0050] Further, in the process of training the style information representation generator, the structure information representation generator, each performance parameter predictor, and the classification network, for each input sample, the style information representation generator, the structure information representation generator, each performance parameter predictor, and the classification network will calculate the deviation between the output value and the true value corresponding to the sample. The deviation will be conducted to different neural network layers of the style information representation generator, the structure information representation generator, each performance parameter predictor, and the classification network through back propagation, so as to adjust the first learning parameters, the second learning parameters, the third learning parameters, and the fourth learning parameters of these neural networks, until the deviation between the output value and the true value for all training samples is minimized.

[0051] In summary, compared with the existing prediction technology of clothing attributes based on image recognition, the present application introduces design data such as knitting structure data and yarn data that cannot be identified by vision, and represents the knitting structure of the knitted product according to the knitting type in different regions, so that the performance parameters such as shrinkage rate, weight, elastic recovery rate, moisture absorption and perspiration coefficient, softness, shear coefficient, deformation coefficient, etc. that are irrelevant / weakly related to visual features such as style, size, shape, etc. can be more accurately predicted, to meet the demand of designers for multi-dimensional performance parameters of knitted products and the demand for professional-level prediction accuracy.

[0052] Please continue to refer to Figure 1After the prediction values of at least one performance parameter closely related to product quality and comfort, such as size, shrinkage, weight, elasticity, sweat coefficient, softness, shear coefficient, deformation coefficient, etc. of the knitted product are determined, the design device of the knitted product can import the performance parameter prediction results output by the model into the 3D proofing software together with the DIS BMP picture design document, so that the 3D software simulates the proofing according to the prediction values of the performance parameters, thereby improving the accuracy of the 3D simulation proofing. In this way, the designer can judge whether the proofing effect of the product meets the expected design requirements based on the simulation proofing effect generated by the 3D software. If the simulation proofing effect generated by the 3D software does not meet the expected product design requirements, the designer can continue to modify the design data of the knitted product, such as style, size, shape, knitting structure type at each position, yarn material at each position, and yarn count at each position, using the design program, and repeat the steps of performance parameter prediction and 3D simulation proofing until the simulation proofing effect generated by the 3D software meets the expected product design requirements. It can be understood that the specific principle of using 3D proofing software for simulation proofing is not related to the technical improvement of the present application and will not be described here.

[0053] Further, in some embodiments, in response to the performance of the knitted product meeting the expected judgment result, the designer can make an entity sample of the knitted product according to the current design data, and test the sample to obtain the actual measurement data of the sample. In this way, the designer can verify whether the performance of the knitted product actually meets the expectation according to the actual measurement data. If the actual performance parameters of the sample actually meet the expected product design requirements, the designer can complete the product design. Otherwise, if the actual performance of the sample does not meet the expected verification result, the designer can continue to modify the design data of the knitted product, such as style, size, shape, knitting structure type at each position, yarn material at each position, and yarn count at each position, using the design program, and repeat the steps of performance parameter prediction, 3D simulation proofing, entity proofing, and sample measurement until the actual effect of the produced sample meets the expected product design requirements.

[0054] Further, in response to the performance of the knitted product not meeting the expected verification result, the design device of the knitted product can also preferably record the current design data of the knitted product and the actual measurement data corresponding to the sample as historical error data. When the recorded historical error data reaches a preset number threshold, the technical personnel can also use these historical error data as data samples to retrain and correct each performance predictor of the knitted product, so as to further improve the accuracy of performance prediction.

[0055] Compared with the current situation that a designer needs to repeat the design- trial production-verification process for at least 15 days to finally complete the design finalization of a product, the performance prediction technology of the knitted product provided by the first, third and fifth aspects of the present application, the design method of the knitted product, the device and the storage medium provided by the second, fourth and sixth aspects of the present application can greatly reduce the workload of product trial production and performance parameter verification. By accurately predicting at least one performance parameter closely related to product quality and comfort such as product size, shrinkage, weight, elasticity, sweat coefficient, softness, shear coefficient and deformation coefficient, and using 3D simulation proofing to replace product trial production and actual measurement process, the designer only needs to perform 1-2 times of product trial production proofing to complete the final design finalization of a product, thereby shortening the design cycle by more than half the time.

[0056] Although the above methods are illustrated and described as a series of acts, it will be understood and appreciated that the methods are not limited by the order of acts, as some acts may, in accordance with one or more embodiments, occur simultaneously or in different order than shown and described herein, or may be omitted altogether, depending on the implementation.

[0057] Those skilled in the art will understand that information, signals, and data can be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0058] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

[0059] Although the performance prediction apparatus and design apparatus described in the above embodiments can be realized by a combination of software and hardware, it is understood that the performance prediction apparatus and design apparatus can be implemented in software or hardware alone. For hardware implementation, the performance prediction apparatus and design apparatus can be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described above and a selection of these. For software implementation, the performance prediction apparatus and design apparatus can be implemented with separate software modules, such as procedures and functions, each of which perform one or more functions and operations described herein when executed by a general-purpose chip.

[0060] The various illustrative logical blocks, circuits, and modules described in connection with the embodiments disclosed herein can be implemented or performed with a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor can be a microprocessor, but in the alternative, the processor can be any conventional processor, controller, microcontroller, or state machine. A processor can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0061] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

[0062] In one or more exemplary embodiments, the functions described can be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions can be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media can be any available media that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk and disc, as used herein, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0063] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples described herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of performance prediction of a knitted product, characterized by, comprising the following steps: obtaining overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product, wherein the overall style data in multiple dimensions includes visual data of the knitted product, yarn data of at least one yarn involved in the knitted product, and / or device data of at least one knitting device involved in the knitted product; dividing the knitted product into multiple regions according to the knitting structure data, wherein each region adopts a knitting structure; each of the regions is respectively disassembled into a plurality of pixel points according to position, to form a first multi-dimensional vector s ij characterizing the knitting structure type of each of the pixel points, and forming a second multi-dimensional vector y ij characterizing the yarn data of each of the pixel points; According to the first multi-dimensional vector s ij and the second multi-dimensional vector y ij , determine the structure information vector of each pixel point t ij , and combine the structure information vectors of each pixel point according to the position t ij to determine the structure information representation of the knitted product in the area T ; stitching the visual data, the yarn data of the at least one yarn, and the device data to determine a style information representation of the knit product D ; The structural information representation is assembled. T and the style information representation D To determine the performance prediction initialization characterization of the knitted product. P 0, and input it to at least one pre-trained performance parameter predictor, and initialize the performance prediction representation according to the corresponding third learning parameters. P 0 Perform at least one convolution operation to determine a performance prediction representation that includes at least one-dimensional latent features. P ;as well as According to the performance prediction characterization P at least one performance parameter corresponding to the performance parameter predictor is determined.

2. The performance prediction method of claim 1, wherein, said first multi-dimensional vector s ij The step of characterizing the type of knit structure of each of the pixels comprises: characterizing the knitting structure type of each pixel point with a first multi-dimensional one-hot vector.

3. The performance prediction method of claim 2, wherein, The yarn data of each of the positions indicates a yarn material and a yarn count of the corresponding position, and the second multi-dimensional vector y ij The step of characterizing the yarn data of each of the pixels includes: characterizing the yarn material of each pixel point with a second multi-dimensional one-hot vector; characterizing the yarn count of each pixel point with second normalized data; and concatenating the second multi-dimensional one-hot vector and the second normalized data to determine the second multi-dimensional vector of yarn data characterizing each of the pixels y ij .

4. The performance prediction method of claim 3, wherein, said structure information vector of each said pixel point is combined according to position t ij to determine a structure information representation of the knitted product in the area T comprises: combining the structural information vectors of each of the pixels according to the position t ij to determine the structural initialization representation of each of the regions of the knitted product T 0; and initializing the structure representation T 0inputting a pre-trained structure information representation, and determining the structure initialization representation according to first learning parameters therein T 0performing at least one convolution operation to determine a structure information representation comprising at least one-dimensional latent features T .

5. The performance prediction method of claim 1, wherein, the visual data includes at least one of style data, size data, and shape data of the knitted product.

6. The performance prediction method of claim 5, wherein, the style data is characterized by a third multi-dimensional one-hot vector, and / or the size data is characterized by fourth normalized data, and / or the shape data is characterized by a fifth multi-dimensional one-hot vector, and / or the yarn data of the at least one yarn includes yarn material data and / or yarn attribute data, and is characterized by a sixth multi-dimensional vector / matrix, and / or the device data includes knitting device type data and / or needle cylinder size data, and is characterized by a seventh multi-dimensional vector / matrix.

7. The performance prediction method of claim 6, wherein, the step of stitching the visual data, the yarn data of the at least one yarn, and the device data to determine a style information representation of the knitted product D includes: concatenating the third multi-dimensional one-hot vector, the fourth normalized data, the fifth multi-dimensional one-hot vector, the sixth multi-dimensional vector, and / or the seventh multi-dimensional vector to determine a style initialization representation of the knit product D 0; and initializing the style representation D 0inputting a pre-trained style information representation, and determining the style initialization representation according to second learning parameters therein D 0performing at least one convolution operation to determine a style information representation comprising at least one-dimensional latent features D .

8. The performance prediction method of claim 1, wherein, The at least one performance parameter includes at least one of product size, shrinkage rate, weight, elastic recovery rate, moisture absorption and perspiration coefficient, softness, shear coefficient, and deformation coefficient of the knitted product.

9. A method of designing a knitted product, characterized by, comprising the following steps: obtaining design data of the knitted product, wherein the design data includes overall style data of the knitted product in multiple dimensions, and knitting structure data and yarn data of multiple positions of the knitted product; implementing the performance prediction method of the knitted product according to any one of claims 1-8 to determine at least one performance parameter of the knitted product; determining whether the performance of the knitted product meets the expectation according to the at least one performance parameter; and in response to the determination result that the performance of the knitted product does not meet the expectation, modifying the design data of the knitted product.

10. The design method of claim 9, wherein, further comprising the following steps: in response to the determination result that the performance of the knitted product meets the expectation, manufacturing a sample of the knitted product according to the design data; testing the sample to obtain measured data of the sample; verifying whether the performance of the knitted product meets the expectation according to the measured data; in response to the verification result that the performance of the knitted product does not meet the expectation, recording the design data of the knitted product and the measured data of the sample as a data sample for correcting the performance predictor of the knitted product, and / or modifying the design data of the knitted product.

11. An apparatus for predicting the performance of a knitted product, characterized by comprising: a memory; and a processor connected to the memory and configured to implement the performance prediction method of the knitted product according to any one of claims 1-8.

12. A design device for a knitted product, characterized by comprising: a memory; and a processor connected to the memory and configured to implement the design method of the knitted product according to claim 9 or 10.

13. A computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions, when executed by a processor, implement a method for predicting the performance of a knitted product as claimed in any one of claims 1 to 8.

14. A computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions, when executed by a processor, implement a method for designing a knitted product as claimed in claim 9 or 10.

Images