TWI792035B - Material recommendation system and material recommendation method for making products - Google Patents

Abstract

A material recommendation system and a material recommendation method for making products are provided, which use an analysis module to analyze at least one image to generate reference information, and then a recommendation module receives the reference information to provide target information corresponding to the reference information. By analyzing the image, it is quickly to provide target information of suitable materials, which can be referred with and greatly accelerate the timeline of product development.

Description

Material recommendation system and material recommendation method for making products

The present invention relates to a material recommendation system, in particular to a material recommendation system and a material recommendation method for selecting suitable materials by means of artificial intelligence.

As humans continue to develop new products to improve the quality of life and promote social progress, the development of new products not only involves technical aspects, but also requires appropriate materials for production. At present, when product developers are looking for appropriate materials, they need to find different material suppliers for each component, and a single component often has different material suppliers due to its specification requirements, so it takes a lot of time to complete the material combination of all components .

Furthermore, if the product needs to be customized, such as athletes of various sports, due to different body shapes and body extension conditions, the specifications (such as stretch rate) required by each part are very different (for example, runners focus on leg stretch rate) , baseball pitchers focus on arm stretch rate), so the material combination (such as water resistance, flexibility, etc.) required by the product (such as smart watch, clothes, etc.) It is not easy to find possible combinations of various materials when selecting materials.

Therefore, how to overcome the various deficiencies of the above-mentioned conventional technologies has become a difficult problem to be overcome urgently in the industry.

In order to solve the various problems of the above-mentioned conventional technologies, the present invention provides a material recommendation system for making products, which includes: a host terminal, which includes an analysis module of a loading learning mechanism and a recommendation module of a loading prediction mechanism. The analysis module is used to analyze at least one image to generate reference information, and the recommendation module is communicatively connected to the analysis module to receive the reference information and provide target information corresponding to the reference information; and the operation terminal is communicatively connected The host end includes a user interface for controlling the host end.

The present invention also provides a material recommendation method for making products, which includes: analyzing at least one image by an analysis module of a loading learning mechanism to generate reference information; and analyzing the image by a recommendation module of a loading prediction mechanism Reference information to provide target information corresponding to the reference information.

It can be seen from the above that the material recommendation system and material recommendation method for making products of the present invention mainly provide target information including applicable materials quickly by analyzing images. The material recommendation system for the production of products can quickly obtain suggestions for material selection to quickly complete the material combination of all components, thus greatly speeding up the product development process.

Furthermore, for the production of customized products, such as making smart watches for athletes of various sports, by using the material recommendation system and material recommendation method for making products of the present invention, product developers can easily obtain the material combinations required by various athletes .

1: Material recommendation system

1a: Host side

1b: Operation terminal

10: Analysis module

100: The first machine learning module

11:Recommended modules

111: filter

112: The second machine learning module

113: Auxiliary

12: Database

123: Human Body Parts

124: skin stretch rate

130: Convolutional Neural Networks

131: public information

132: Known information

150: human body

170: Convolutional Neural Networks

171: public information

172: Known information

80: User interface

81:Sharing mechanism of rules of experience

82:Data Feedback Mechanism

9: Electronic device

90: Material information

90': Advanced information

91: Learning mechanism

92:Prediction mechanism

A1, A2: Image

B: Comparison table of male and female average stretch rate

P0: Video

P0’,P0”: photo

P1,P2,P3: list

P1', P2', P3': list

S10~S16: Steps

S20~S27: steps

S270~S276: steps

S361~S368: Steps

S40~S46: steps

S42': step

S50: Steps

S501~505: steps

T1~T8: parts

FIG. 1 is a schematic diagram of the operational framework of the material recommendation system for making products according to the present invention.

FIG. 1A is a schematic diagram of the operation flow of the learning mechanism in FIG. 1 .

FIG. 1B is a schematic diagram of the operation flow of the prediction mechanism in FIG. 1 .

Fig. 1B' is a schematic diagram of the operation flow of the advanced forecasting operation in Fig. 1B.

FIG. 2 is a schematic diagram of the functional architecture of the host terminal of the material recommendation system for making products according to the present invention.

FIG. 2A is a schematic diagram of the operation flow of the analysis module in FIG. 2 .

FIG. 2B is a schematic diagram of the machine learning process of the analysis module in FIG. 2 .

FIG. 2C is a schematic diagram of public data used by the analysis module of FIG. 2 in machine learning.

FIG. 2D is a schematic diagram of known data used by the analysis module of FIG. 2 in machine learning.

FIG. 3A is a schematic diagram of the operation flow of the recommendation module in FIG. 2 .

FIG. 3B is a schematic diagram of the machine learning process of the recommendation module in FIG. 2 .

FIG. 4 is a schematic flowchart of a method for recommending materials for making products according to the present invention.

FIG. 5 is a schematic flow chart of the auxiliary operation in FIG. 4 .

FIG. 6A is a schematic diagram of one embodiment of the process of recommending materials using the material recommending system for making products of the present invention.

FIG. 6B is a schematic diagram of another embodiment of recommending materials using the material recommending system for making products of the present invention.

The implementation of the present invention is described below through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification.

It should be noted that the structures, proportions, sizes, etc. shown in the drawings attached to this specification are only used to match the content disclosed in the specification, for the understanding and reading of those familiar with this technology, and are not used to limit the implementation of the present invention Therefore, it has no technical substantive meaning. Any modification of structure, change of proportional relationship or adjustment of size shall still fall within the scope of this invention without affecting the effect and purpose of the present invention. The technical content disclosed by the invention must be within the scope covered. At the same time, terms such as "first", "second", "upper", "lower" and "one" quoted in this specification are only for the convenience of description, and are not used to limit the scope of the present invention. The scope of implementation and the change or adjustment of its relative relationship shall also be regarded as the scope of implementation of the present invention without substantial change in technical content.

FIG. 1 is a schematic diagram of the structure of a product-making material recommendation system 1 of the present invention. As shown in Figure 1, the material recommendation system 1 described includes a host end 1a and an operation end 1b, the host end 1a is operated by an electronic device 9, and the operation end 1b is a client side (client side) , which controls the host terminal 1a through a user interface 80 to acquire target information of the target object.

In this embodiment, the electronic device 9 is a computer host or a cloud device, which can communicate with different user interfaces (user interface) 80 (such as network mode), and the user interface 80 is configured on a home computer, Laptops, smartphones, tablets or other appropriate 3C products.

Furthermore, the target object is, for example, clothing or a wristband and other wearing items.

The host end 1a has a database 12 , a learning mechanism 91 and a prediction mechanism (predictor) 92 .

The database 12 can be used to store material data 90, advanced data 90' (such as measurement methods (testing methods) and their results (results), media (vehicles) or others) or other supplementary data as required, as the learning source of the learning mechanism (that is, the input of the learning mechanism). For example, the material information 90 includes related information of various materials, such as flexible materials, flexible materials, waterproof materials, breathable materials Materials, conductive materials or other materials and their properties (properties) and sources, etc. Specifically, various data sets can be designed according to the type of material required, such as flexible material data sets, waterproof material data sets, breathable material data sets, conductive material data sets or other data sets required by the target object wait.

The learning mechanism 91 can be an artificial intelligence training engine (AI training engine), so as to operate the prediction mechanism 92 based on the output of the learning mechanism. Specifically, as shown in FIG. 1A , the training process of the learning mechanism 91 is as follows.

In step S10 , a collection operation is performed to receive relevant data from the database 12 .

In step S11, a preparatory operation is performed to pre-process (such as eliminating, classifying, formatting or other actions) the collected data.

In step S12, a calculation operation is performed to process the prepared data using a multicollinearity calculation.

In step S13, a remove operation is performed to remove the restriction of multicollinearity together with the data that has completed the multicollinearity and the data from the empirical law sharing mechanism 81 of the operation terminal 1b.

In step S14, a calculation operation is performed to use a linear (linear) algorithm or a nonlinear (non-linear) algorithm to learn the data from which the multicollinearity has been removed, so as to generate new data.

In step S15, a judging operation is performed to judge whether the performance of the training through the calculation operation is good or not.

In step S16 , if the training result is good, a construction operation is performed to form a recommendation rule for input into the prediction mechanism 92 . Otherwise, then get back to the preparatory work of step S11 and study again.

The forecasting mechanism 92 is used to perform forecasting operations, so as to recommend (for example, network transmission) the forecasting results to the operation terminal 1b. Specifically, as shown in FIG. 1B , the prediction process of the prediction mechanism 92 is as follows.

In step S20, an extraction operation is performed, which receives the recommendation principle of the learning mechanism 91 and the demand information from the operation terminal 1b, wherein the operation terminal 1b inputs (imports) the demand information through the user interface 80 to the electronic device 9, and the demand information includes vehicles, targets, score criteria or others.

In step S21, a search operation is performed to obtain all materials in the database matching the requirement information by means of calculation.

In step S22, a prediction operation is performed, which is based on the materials obtained in the search operation, to predict the relevant material combination of the required characteristics of the target object of the demand information.

In step S23, a calculation operation is performed, which calculates the recommendation score of each material based on each material combination obtained from the prediction operation.

In step S24, a judging operation is performed to judge whether all the materials required by the object of the demand information have been calculated.

In step S25, if the judgment of step S24 is "Yes", the allocation operation is performed to select materials that meet the criteria of the source as an accurate combination, if the suggested score is greater than the criteria of the source. Otherwise, then return to the search operation of step S21 to search again.

In step S26, at least one accurate combination is sorted (ranking) as the prediction result or target information, and presented (show) on the user interface 80 for reference by the operation terminal 1b, wherein the prediction result or The target information system includes the material data and its source. For example, a single flexible material can come from a single source or multiple sources, and the source is a supplier (such as a manufacturer or a person) or a manufacturer (such as a manufacturer or a person).

In step S27, if the exact combination is not generated in step S25, an advanced prediction (backward predictor) operation (or auxiliary operation) is performed. Specifically, as shown in FIG. 1B', the operation flow of the advanced forecasting operation S27 is as follows.

In step S270, the characteristics and medium of the object of the demand information are set.

In step S271, classify each characteristic range to form a plurality of intervals for use as a searching space, so as to create a searching space for the characteristic.

In step S272, an optimization algorithm is selected, for example, from Grid Search, Random Search, Bayesian optimization, Evolutionary Algorithm ), reinforcement learning (Reinforcement Learning), or other appropriate methods.

In step S273, sampling is performed in each of the search spaces with an optimization algorithm.

In step S274, according to the sampling result, the material property of the object is calculated.

In step S275, a comparison is performed to determine whether there is a material meeting the source evaluation.

In step S276, if the comparison result is "Yes", similar combination information or optimization (properly) suggestions that meet the evaluation of the source will be generated, and the similar combination information (or optimization suggestions) will be returned (return) to displayed on the user interface 80 . Otherwise, return to step S273 for re-sampling.

The operation terminal 1b includes the user interface 80 , a rule of thumb sharing mechanism 81 and a data feedback mechanism 82 .

In this embodiment, the user interface 80 is a graphical user interface (Graphical User Interface, referred to as GUI), in order to facilitate the operation terminal 1b to use the material recommendation system 1, and the rule of thumb sharing mechanism 81 and data feedback mechanism 82 are both Corresponding operation options are configured on the user interface 80 .

Furthermore, the prediction mechanism 92 of the host terminal 1a will present the prediction result or target information on the user interface 80, and the operation terminal 1b can use the rule of thumb sharing mechanism 81 to input relevant materials and data applied by the user into The learning mechanism 91 of the host terminal 1a can, through the data feedback mechanism 82, supplement the data of actually using the prediction result or target information to make the target object to the database 12 of the host terminal 1a, so as to strengthen the learning mechanism The learning effect of 91 is conducive to the prediction effect of the prediction mechanism 92.

FIG. 2 is a schematic diagram of the functional structure of the host terminal 1a of the product-making material recommendation system 1 of the present invention. As shown in Figure 2, the host terminal 1a described includes: a database 12 as described in Figure 1, an analysis module 10 equipped with the learning mechanism 91, and a recommendation module equipped with the prediction mechanism 92 Group 11.

The analysis module 10 includes an image recognition unit (not shown), which is used to analyze at least one image to generate reference information based on the recommendation principle in step S16 shown in FIG. 1A .

In this embodiment, the image is in the form of a video or a photo. Preferably, the image at least needs to include different poses of the previous and subsequent stages. For example, a single video or multiple photos are in continuous pose form.

Furthermore, the content of the image includes at least a part of the outline of the human body, such as shoulders, front thighs, knees, chest, back, wrists, elbows or other parts. It should be understood that the content of the image may also include at least a part of other animals or objects.

Also, the reference information includes stretch rate, such as the skin stretch rate of human body parts (the value ranges from 0% to 120%) or the physical stretch rate of the mechanism. For example, the analysis module 10 can perform prediction based on the images A1 and A2 to generate human body parts 123 (such as shoulders, front thighs, knees, chest, back, wrists, elbows or other parts, etc.) in the images A1 and A2. ) skin stretch rate 124, the operation process shown in Figure 2A. Specifically, the analysis module 10 includes a first machine learning model 100 that operates the learning mechanism, such as a support vector machine (Support Vector Machine, referred to as SVM) model, a convolutional neural network (Convolutional Neural Networks, referred to as CNN) algorithm model, random forest (Random Forest) model, nearest neighbor (k-Nearest Neighbors, referred to as KNN) algorithm or other artificial intelligence (Artificial Intelligence, referred to as AI) model, to train (the training process shown in Figure 2B, with Take the convolutional neural network 130 as an example), in addition to using the database 12 for training, it can also use public data 131 (such as human historical action images or the fiber engineering journal shown in Figure 2C in 1983 AD Vol. 36. The comparison table B of the average stretch rate of men and women disclosed in No. 6) or other known data 132 (such as the stretch rate of each part obtained by actually measuring the demonstration personnel of various postures in various postures, such as One of the postures shown in Figure 2D shows the stretch rate of each position obtained) training (the input of the rule of thumb sharing mechanism 81 of step S13 as shown in Figure 1A), so that when inputting a film or one (or multiple) For images of the same human body part, the analysis module 10 can predict the skin stretch rate of the human body part based on the recommendation principle, wherein the stretch rates of T1-T8 of each part of the human body 150 shown in FIG. 2D are shown in Table 1 below Show:

The recommendation module 11 is communicatively connected to the analysis module 10 to receive the reference information, and executes the operation process shown in FIG. The target information of the reference information (such as the prediction result of the sorting operation in step S26 ) is used for subsequent production of target objects.

Specifically, in this embodiment, the recommendation module 11 can be configured with a filter 111 connected to the database 12, so that the filter 111 can select the material data of the required target information from the database 12 (step S20~step S25 shown in FIG. 1B ), so that the recommendation module 11 can provide an accurate combination of the material data according to the requirement of the target object (step S26 shown in FIG. 1B ). For example, for a baseball pitching action, since the elongation rate of the chest is different from the elongation rate of the wrist, the flexible materials used in the two parts are different, so the filter 111 will select the two parts from the database 12. The flexible material required by the parts can be used to make clothes that meet the stretch specifications.

Furthermore, because the material information in the database 12 may be insufficient, the filter 111 cannot select flexible materials that meet the tensile specifications, so the recommendation module 11 can be equipped with an auxiliary device 113 that is connected to the database 12, Calculate the flexible material that approximates the tensile specification or select the flexible material that approximates the tensile specification from the database 12 (step S27 shown in Figure 1B and the step shown in Figure 1B' S270~step S276). It should be understood that the filter 111 and/or the auxiliary device 113 can also select the air-permeable material, conductive material or other specification materials required by each part from the database 12 according to the requirements, so as to produce a material that meets the specification requirements. the target object.

Moreover, the recommendation module 11 may also include a second machine learning model 112 connecting the database 12 and the filter 111, such as the least absolute shrinkage and selection operator (Lasso for short) model , support vector machine (SVM) model, convolutional neural network (CNN) calculus model, random forest model, nearest neighbor (KNN) algorithm or other artificial intelligence (AI) models, which can utilize the open Data 171 (such as various flexible materials and their basic characteristics) or other known data 172 (actual measurement of various materials and their basic characteristics) for training (the process shown in Figure 3B, to use convolutional neural network 170 as an example), when the input (as shown in Figure 1B When the operation end of step S20 inputs) skin stretch rate or other specification requirements, the filter 111 can select the required material data through the second machine learning model 112 .

In addition, by storing the sources of various material data in the database 12, the recommendation module 11 can not only provide an accurate combination or a similar combination of material data required by the target object, but can further provide the source of each material data.

It should be understood that if the analysis module 10 uses the first machine learning model 100 for training, the reference information may include other specification conditions, such as hardness, etc., not limited to elongation, so the recommendation module 11 may Provide the target information (such as the exact combination or similar combination of material data required by the target) for various specification conditions (such as biocompatibility, anti-sweat corrosion, resistance change rate, etc.).

Fig. 4 is a flow chart of the method for recommending materials for making products according to the present invention. As shown in FIG. 4 , the material recommendation method works in conjunction with the material recommendation system 1 . In this embodiment, the target object is sports clothing, so the material recommendation method is used to query materials required by the sports clothing.

In step S40, at least one image, such as a video or one to five images (photos), is provided. In this embodiment, the user can upload the image to the electronic device 9 through the user interface 80 .

In step S41, the analysis module 10 analyzes the image. In this embodiment, the image is input into the first machine learning model 100 to perform image recognition and stretch state analysis.

In step S42, the analysis module 10 generates reference information based on the recommendation principle. In this embodiment, after the image recognition and stretch state analysis of the first machine learning model 100, the first machine learning model 100 outputs a prediction result, which includes the skin stretch rate (such as a stretch rate of 20% ) or other reference information for specification requirements.

In step S43 , the reference information is input into the recommendation module 11 to perform the prediction mechanism 92 . In this embodiment, other insertion methods can be used again (step S20 as shown in Figure 1B) Manual input of the operation terminal 1b) input another reference information (such as waterproof, anti-corrosion and other specification requirements) into the recommendation module 11, as shown in step S42', so that the recommendation module 11 receives multiple sets of reference information .

In step S44, the recommendation module 11 performs screening according to the reference information (step S21-step S25 shown in FIG. 1B). In this embodiment, the reference information is input into the second machine learning model 112 , so that the second machine learning model 112 outputs a prediction result according to the database 12 after analyzing the reference information.

In step S45, the recommendation module 11 provides target information corresponding to the reference information (step S26 shown in FIG. 1B). In this embodiment, if the prediction result shows that the database 12 has an accurate combination of material data required by the reference information (such as a stretch rate of 40%, which is greater than the stretch rate of the reference information of 20%, that is 40>20), then the filter 111 can select the exact combination of required material data from the database 12, and the target information can present the source of the material data, such as the information about each part of the sportswear. Manufacturer (or material supplier). It should be understood that this exact combination means that all materials meet specification requirements.

In step S46, the target information is displayed on the screen of the user interface 80 for the user's reference.

On the other hand, in step S50, if the prediction result is unable to provide the exact combination of material data required to meet the reference information from the database 12 (such as only data with a stretch rate of 5% in the database 12 , which is less than 20% of the elongation rate of the reference information, that is, 5<20), the prediction result will be input into the assistant 113 to perform auxiliary operations (or as described in step S27 as shown in FIG. 1B ) Step prediction operation), so that the assistant 113 provides a similar combination of required material data (such as stretch rate 18%, which is close to the stretch rate 20% of the reference information), as another target information, and make the Another target information is displayed on the screen of the user interface 80, as shown in step S46. It should be understood that the close combination means that at least one material does not meet the specification requirement.

In this embodiment, the similar combination of the material data can be provided to the material developer, so that they can consider and develop relevant materials for the operation terminal 1b to use the data feedback mechanism 82 to make up for the shortage of the database 12 .

Furthermore, the operation process of the assisting device 113 performing the assisting operation is shown in FIG. 5 and will be described in detail below.

In step S501, a classification operation is performed. In this embodiment, the characteristic ranges of the materials are classified to form a plurality of intervals, which are used as search spaces, as shown in step S270~step S271 shown in FIG. 1B'. For example, classify the properties of the materials listed in this reference into the flexibility zone, water resistance zone, and other zones.

In step S502, an optimization operation is performed. In this embodiment, an optimization algorithm is used to sample in each of the search spaces, as shown in step S272 to step S273 shown in FIG. 1B'. For example, the optimization algorithm includes grid search (Grid Search), random search (Random Search), evolutionary algorithm (Evolutionary Algorithm), reinforcement learning (Reinforcement Learning) or other appropriate methods to obtain the Select relevant information from the water resistance interval, water resistance interval and other intervals, such as videos, photos, literature or other public information about materials.

In step S503, prediction is made according to the sampling result. In this embodiment, artificial intelligence (AI) calculation method or other calculation methods are used to predict the elongation rate (and other specifications), as in step S274 shown in FIG. 1B'.

In step S504, the prediction results output by the first machine learning model 100 are analyzed to determine whether there is a material with a corresponding stretch rate (and other specifications), as in step S275 shown in FIG. 1B'.

In step S505, if the analysis result is a material with a corresponding elongation (and other specifications), an optimization suggestion is provided as the other target information, as in step S276 shown in FIG. 1B'.

On the other hand, if there is no material corresponding to the elongation ratio (and other specifications) in the analysis result, the optimization calculation operation S502 is performed again.

Therefore, the material recommendation system 1 and material recommendation method for making products of the present invention use the stretching image of the same part of the human body to predict the skin stretch rate with the analysis module 10, and match it with other items of the target object (such as sports clothing). According to the specification requirements, the recommendation module 11 is used to obtain an accurate combination containing the material data. If there is no accurate combination of the material data, a similar combination of the material data is proposed as an auxiliary suggestion. In other words, the material recommendation system 1 and the material recommendation method search the existing data, and if there is no exact combination of suitable materials in the existing data, the assistant 113 can predict a similar combination of the material data.

FIG. 6A is a flow chart of the first embodiment of the actual operation of the material recommendation system 1 applied to the material recommendation method. In this embodiment, the user enters the host terminal 1a through the user interface 80 of the operation terminal 1b to inquire about the materials required for making sportswear containing electronic components.

As shown in FIG. 6A , the user uploads an image of a person (such as a pitching video P0 ) to the electronic device 9 through the user interface 80 . Next, image recognition is performed through the analysis module 10 to obtain stretching ratios of various human body parts, as shown in the list P1. Afterwards, the analysis module 10 inputs the reference information about the elongation rate into the recommendation module 11, and the user can input the required specifications ( As shown in the list P2) into the recommendation module 11, so that the recommendation module 11 filters out the target information about the sports clothing, such as the accurate material of each component (such as substrate, wire, packaging material) of the electronic component ( Such as material No. 015) and its source (such as Company A). Finally, the recommendation module 11 outputs the target information (as shown in the list P3 ) to be displayed on the user interface 80 for the user's reference.

In the second embodiment shown in FIG. 6B , the user uploads images with people (such as multiple swimming photos P0', P0") to the electronic device 9 through the user interface 80. Then, through the analysis module 10 Carry out image recognition to obtain the stretch ratio of each body part, as shown in the list P1'. After that, the The analysis module 10 inputs the reference information about the elongation rate into the recommendation module 11, and the user can input the required specifications of the sports clothing (target object) that includes electronic components from the user interface 80 (such as list P2 ') to the recommendation module 11, so that the recommendation module 11 performs screening. When there is no accurate combination of suitable materials in the material screening range, the auxiliary device 113 is used to perform auxiliary operations to propose optimization suggestions, such as similar materials for wires (if there is a material supplier, the material supplier will be displayed in the target information above; if there is no material supplier, the user can provide it to the material supplier for development and customization). Finally, the recommendation module 11 outputs target information (as shown in the list P3') to be displayed on the user interface 80 for the user's reference.

To sum up, the material recommendation system and material recommendation method for making products of the present invention can provide target information by analyzing images, so the material recommendation system of the present invention can quickly obtain suggestions for material selection to greatly reduce The time to complete the combination of materials for all components can greatly accelerate the time of product development.

Furthermore, for the production of customized products, the material combination required for various customized products can be easily obtained through the material recommendation system of the present invention.

The above-mentioned embodiments are only illustrative to illustrate the principles and effects of the present invention, and are not intended to limit the present invention. Anyone skilled in the art can modify and change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the rights of the present invention should be listed in the scope of the patent application described later.

1: Material recommendation system

1a: Host side

1b: Operation terminal

12: Database

80: User interface

81:Sharing mechanism of rules of experience

82:Data Feedback Mechanism

9: Electronic device

90: Material information

90': Advanced information

91: Learning mechanism

92:Prediction mechanism

Claims (12)

A material recommendation system for making products, which includes: a host terminal, which includes an analysis module of a loading learning mechanism, a database for storing material data, and a recommendation module of a loading prediction mechanism. The analysis module is for analyzing at least one image to generate reference information including elongation ratio, and the recommendation module is communicatively connected to the analysis module to receive the reference information and cause the recommendation module to select the product from the database The required material data, so that the recommendation module provides target information corresponding to the reference information, and the target information includes the material data and the source of the material data, wherein the analysis module includes a first A machine learning model for inputting the image into the first machine learning model for image recognition and stretch state analysis, and outputting a prediction result including the reference information through the first machine learning model; and operation The terminal is connected to the host terminal through communication, and includes a user interface for controlling the host terminal. The material recommendation system according to Claim 1, wherein the image includes at least a part of the outline of the human body. The material recommendation system according to Claim 1, wherein the analysis module analyzes the plurality of images, and the plurality of images are in the form of continuous poses. The material recommendation system as described in Claim 1, wherein the recommendation module is configured with a filter connected to the database, so that the filter selects the required material data from the database. The material recommendation system as described in claim 1, wherein the recommendation module includes a second machine learning model that operates the prediction mechanism. The material recommendation system as described in claim item 1, wherein the recommendation module is configured with an assistant connected to the database, so that the assistant can calculate similar required material data or select similar ones from the database The required material information is used as another target information. A method for recommending materials for making products, comprising: providing a database for storing material data; analyzing at least one image by an analysis module of a loading learning mechanism to generate reference information including elongation ratio, wherein, The analysis module includes a first machine learning model that operates the learning mechanism, so as to input the image into the first machine learning model to perform image recognition and stretch state analysis, and through the first machine learning model Outputting a prediction result including the reference information; and analyzing the reference information through a recommendation module of a loading forecast mechanism, so that the recommendation module selects the material data required for the product from the database, so that The recommendation module provides target information corresponding to the reference information, and the target information includes the material data and the source of the material data. The method for recommending materials according to claim 7, wherein the image includes at least a part of the outline of the human body. The material recommendation method as described in Claim 7, wherein the plurality of images are analyzed by an analysis module, and the plurality of images are in the form of continuous poses. The material recommendation method as claimed in claim 7, wherein the recommendation module includes a second machine learning model that operates the prediction mechanism. The material recommendation method as described in Claim 7, wherein the recommendation module is configured with a filter connected to the database, so that the filter selects the required material data from the database. The material recommendation method as described in claim item 7, wherein the recommendation module is configured with an auxiliary device connected to the database, so that the auxiliary device can calculate similar required material data or select similar ones from the database The required material information is used as another target information.

Images