CN117727426A - Method and system for constitution identification and matching of medicated diet scheme based on deep learning - Google Patents

Abstract

The application discloses a method and a system for constitution identification and matching of a medicated diet scheme based on deep learning, and relates to the technical field of information technology processing. The method comprises the following steps: the method comprises the steps of construction of an identification model, acquisition and input of images, acquisition of identification values, acquisition of identification results, construction of a collocation model and recommendation of medicated diet. The system is matched with the method. According to the method and the system for identifying and matching the medicated diet scheme based on the deep learning, the constitution is identified through the deep learning of various characterization data, the constitution identification result of the user is obtained, so that the constitution type of the user is known, then the matching of the medicated diet materials is carried out based on the constitution type, the medicated diet suitable for the constitution of the user can be clearly known by the common user, and great convenience is provided for the user.

Description

Method and system for constitution identification and matching of medicated diet scheme based on deep learning

Technical Field

The application relates to the technical field of information technology processing, in particular to a method and a system for constitution identification and matching of a medicated diet scheme based on deep learning.

Background

With the development of social economy and the improvement of living standard, people are paying more attention to diet health. However, different constitutions require attention to the matching of food materials in daily diets, as expressed by the policy of traditional Chinese medicine diet, i.e. "lively medical food". In addition, some constitutions which have influence on the body operation are more required to relieve or even improve the influence of the constitutions on the body in a dietetic therapy mode. However, the medicinal materials are difficult for common people to simply read and clearly, which physique corresponds to which medicinal diet, and the medicinal materials are deeply researched by a strong pathological basis, so that reasonable medicinal diet collocation is difficult to realize for common people. Therefore, the application provides a technology for combining physique and medicated diet scheme collocation.

Disclosure of Invention

The application aims to provide a method and a system for identifying constitution and matching a medicated diet scheme based on deep learning, so as to solve the technical problems in the background technology.

In order to achieve the above purpose, the present application discloses the following technical solutions:

in a first aspect, the present application discloses a method for identifying and matching a medicated diet regimen based on deep learning, the method comprising the steps of:

training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes based on deep learning to obtain a tongue image identification model, training by taking facial features and corresponding face colors as characteristic attributes to obtain a face image identification model, and training by taking eyebrow images and corresponding eye socket colors, eyeball colors and eyeball forms as characteristic attributes to obtain an eye image identification model;

based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model;

inputting the tongue image identification model, the face image identification model and the object image identification model corresponding to identification parameters of a user, wherein the identification parameters comprise: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

the tongue picture identification model performs feature analysis on the tongue picture image based on the input tongue picture image to obtain a tongue picture feature attribute corresponding to the tongue picture image, and performs matching value calculation on the obtained tongue picture feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue picture identification value; the face image identification model performs feature analysis on the face image based on the input face image to obtain corresponding face image feature attributes, and performs matching value calculation on the obtained face image feature attributes and a preset physique result based on a convolutional neural network to obtain a face image identification value; the eye image identification model performs feature analysis on the eye image based on the input eye image to obtain corresponding eye image feature attributes, and performs matching value calculation on the obtained eye image feature attributes and a preset physique result based on a convolutional neural network to obtain an eye image identification value;

performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a phenotype constitution, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a hidden constitution when the correlation description of the tongue region identification value, the facial image identification value and the object identification value meets a preset judgment condition;

inputting the phenotype physique and the invisible physique into the medicated diet collocation model, wherein the medicated diet collocation model recommends medicated diet materials based on the phenotype physique and the type of physique in the invisible physique.

Preferably, the acquiring of the authentication parameter further includes: and carrying out color averaging and normalization processing on the tongue region image, the face image and the eyebrow image.

Preferably, the correlation description specifically includes:

defining a constitution of the same type corresponding to the tongue region discrimination value, the facial image discrimination value and the eye image discrimination value as a constitution to be determined, and calculating a correlation coefficient Ass of the constitution to be determined in the tongue region image, the facial image and the eye-brow image _(A,B,C) ；

The obtained correlation coefficient Ass _(A,B,C) And a preset coefficient threshold Ass _Min Comparing;

when Ass _(A,B,C) ≥Ass _Min When the correlation is described as: the constitution to be determined has correlation among the tongue region image, the face image and the eyebrow image;

when Ass _(A,B,C) ＜Ass _Min When the correlation is described as: the constitution to be determined has no correlation among the tongue region image, the face image and the eyebrow image.

Preferably, the correlation coefficient is calculated by the following formula:

wherein A is the median of the matching values of the constitution to be determined in n times of output through the tongue picture identification model, B is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model, C is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model,for the average value of the matching values output by the tongue picture identification model for the undetermined physique several times,/I>For said constitution to be determined, an average value of the matching values outputted several times by said facial image discrimination model,/->And (3) outputting an average value of matching values for the undetermined physique through the object identification model for a plurality of times.

Preferably, the judgment condition is: and identifying whether the obtained physique has correlation in the tongue region image, the face image and the eyebrow image.

Preferably, when the correlation description of the tongue region identification value, the face image identification value and the eye image identification value does not satisfy a preset judgment condition, one or more of the tongue region image, the face image and the eye image are acquired again, the corresponding tongue region identification value, face image identification value or eye image identification value is acquired again, and then the correlation description is performed.

Preferably, when the repeated relevance description does not meet the preset judging condition, each tongue area identification value, the facial image identification value or the physique type corresponding to the target image identification value is respectively input into the medicated diet collocation model to recommend medicated diet materials.

Preferably, the method for identifying and matching the medical diet scheme based on the constitution of deep learning further comprises the following steps:

classifying the recommended medicated diet materials according to the corresponding physique types, and grading the favorability of the medicated diet materials on the physique according to the pharmacology of the traditional Chinese medicinal materials;

sequentially arranging the obtained phenotype physique and the obtained stealth physique according to the tongue region identification value, the facial image identification value and the object image identification value to obtain a physique sequence Seq _CON ，Seq _CON = { constitution 1, constitution 2,., constitution N };

calculate the medicated diet value CP, cp=k ₁ *M _m1.1 +K ₂ *M _m2.1 +...+K _N *M _mN.1 Wherein M is _mN.1 A medicinal material of a plurality of medicinal materials corresponding to constitution N is shown, K is a medication decision coefficient, and K ₁ ,K ₂ ...K _N E (0, 1), when K _i When the medicine is=0, the corresponding medicine and other medicines are in pharmacological conflict, otherwise, the corresponding medicine is not in pharmacological conflict with other medicines;

and calculating a medicated diet value CP corresponding to all medicinal material combinations in the medicated diet materials, and selecting a group of most optimal medicated diet materials with the largest medicated diet value CP.

Preferably, when calculating the medicated diet value CP, when the medicinal material corresponding to the phenotype body quality has pharmacological conflict with other medicinal materials corresponding to the stealth body quality, adjusting K corresponding to the phenotype body quality in the medicated diet value CP to be 1, and adjusting K corresponding to the stealth body quality in the medicated diet value CP to be 0.

In a second aspect, the present application discloses a system for deep learning-based constitution identification and matching of a medicated diet regimen, which is applicable to the method for deep learning-based constitution identification and matching of a medicated diet regimen as described above, the system for deep learning-based constitution identification and matching of a medicated diet regimen comprising:

the data acquisition module is configured to: an authentication parameter for collecting a user, the authentication parameter comprising: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

the model training module comprises a tongue image identification model unit, a facial image identification model unit, a eye image identification model unit and a medicated diet collocation model unit; the tongue image identification model unit is configured to: based on deep learning, training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes to obtain a tongue image identification model; the face image discrimination model unit is configured to: training by taking facial features and corresponding facial colors as feature attributes based on deep learning to obtain a facial image identification model; the object discrimination model unit is configured to: training by taking an eyebrow image and corresponding eye socket color, eyeball color and eyeball shape as characteristic attributes based on deep learning to obtain an eye image identification model; the medicated diet collocation model unit is configured as follows: based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model; wherein the tongue image identification model is configured to: based on an input tongue region image, carrying out feature analysis on the tongue region image to obtain a tongue image feature attribute corresponding to the tongue region image, and calculating a matching value of the obtained tongue image feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue region identification value; the facial image authentication model is configured to: performing feature analysis on the face image based on the input face image to obtain a corresponding face feature attribute, and performing matching value calculation on the obtained face feature attribute and a preset physique result based on a convolutional neural network to obtain a face identification value; the object identification model is configured to: based on an input eyebrow image, performing feature analysis on the eyebrow image to obtain corresponding object feature attributes, and calculating a matching value of the obtained object feature attributes and a preset physique result based on a convolutional neural network to obtain an object identification value;

a correlation evaluation module configured to: performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, comparing the obtained correlation description with preset judging conditions, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value as a phenotype constitution when the correlation description meets the preset judging conditions, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value as a stealth constitution;

the medicated diet collocation model is configured as follows: and recommending the medicated diet materials by taking the phenotype physique and the stealth physique as input layers and based on the types of the phenotype physique and the stealth physique.

The beneficial effects are that: according to the method and the system for identifying and matching the medicated diet scheme based on the deep learning, the constitution is identified through the deep learning of various characterization data, the constitution identification result of the user is obtained, so that the constitution type of the user is known, then the matching of the medicated diet materials is carried out based on the constitution type, the medicated diet suitable for the constitution of the user can be clearly known by the common user, and great convenience is provided for the user. In addition, the accuracy of the constitution identification result is ensured in the data processing process of constitution identification, and the recommending process of the medicated diet (material) is strict, so that the reliability and the practicability of the technology are ensured.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flow chart of a method for identifying and matching medicated diet schemes based on deep learning according to an embodiment of the present application.

Detailed Description

The following description of the technical solutions in the embodiments of the present application will be clear and complete, and it is obvious that the described embodiments are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.

In this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises the element.

In a first aspect, the present embodiment discloses a method for identifying and matching a medicated diet scheme based on deep learning, as shown in fig. 1, the method comprising the following steps:

construction of an identification model: training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes based on deep learning to obtain a tongue image identification model, training by taking facial features and corresponding face colors as characteristic attributes to obtain a face image identification model, and training by taking eyebrow images and corresponding eye socket colors, eyeball colors and eyeball forms as characteristic attributes to obtain an eye image identification model;

and (3) constructing a collocation model: based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model;

collecting and recording images: inputting the tongue image identification model, the face image identification model and the object image identification model corresponding to identification parameters of a user, wherein the identification parameters comprise: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

acquisition of an authentication value: the tongue picture identification model performs feature analysis on the tongue picture image based on the input tongue picture image to obtain a tongue picture feature attribute corresponding to the tongue picture image, and performs matching value calculation on the obtained tongue picture feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue picture identification value; the face image identification model performs feature analysis on the face image based on the input face image to obtain corresponding face image feature attributes, and performs matching value calculation on the obtained face image feature attributes and a preset physique result based on a convolutional neural network to obtain a face image identification value; the eye image identification model performs feature analysis on the eye image based on the input eye image to obtain corresponding eye image feature attributes, and performs matching value calculation on the obtained eye image feature attributes and a preset physique result based on a convolutional neural network to obtain an eye image identification value;

obtaining an identification result: performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a phenotype constitution, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a hidden constitution when the correlation description of the tongue region identification value, the facial image identification value and the object identification value meets a preset judgment condition;

recommendation of medicated diet: inputting the phenotype physique and the invisible physique into the medicated diet collocation model, wherein the medicated diet collocation model recommends medicated diet materials based on the phenotype physique and the type of physique in the invisible physique.

The constitution identification is an important link of traditional Chinese medicine diagnosis, and is based on constitution, and the health condition of a human body is comprehensively evaluated by methods of observation, auscultation, inquiry, diagnosis and the like, so that the constitution of the human body can be identified by adopting the traditional Chinese medicine technology. According to the method and the device, the physique of the user is comprehensively identified through the representation, the face and the representation of the tongue and other parts of the tongue region and the eyebrow and the representation of the eyebrow, and the physique type of the user with the strongest physical expression is clarified through the acquisition of the phenotype physique and the stealth physique, so that a more clear data basis is provided for recommending the medicated diet materials. Meanwhile, the user can select the constitution type needing to be intervened/adapted by himself, so that customization of the medicated diet materials is realized, and the use experience of the user is improved.

In this embodiment, the obtaining of the authentication parameter further includes: and carrying out color averaging and normalization processing on the tongue region image, the face image and the eyebrow image. The method has the advantages that the calculation pressure of the model in the process of identifying the data can be reduced, and the data processing efficiency and accuracy are improved.

The relevance description may be any of the available technologies, such as: based on the similarity evaluation mode, the tongue area discrimination value, the facial image discrimination value and the numerical result of the object image discrimination value are analyzed, and the corresponding constitution type is clearly expressed, so that the strong constitution type and the like are clearly expressed. However, in this way, it is difficult to control the occurrence of errors, that is, when any one of the tongue region discrimination value, the face discrimination value, and the object discrimination value is in error, it is impossible to ensure the accuracy of the constitution obtained by discrimination.

In this regard, as a preferred implementation of the present embodiment, the correlation description specifically includes:

defining a constitution of the same type corresponding to the tongue region discrimination value, the facial image discrimination value and the eye image discrimination value as a constitution to be determined, and calculating a correlation coefficient Ass of the constitution to be determined in the tongue region image, the facial image and the eye-brow image _(A,B,C) ；

The obtained correlation coefficient Ass _(A,B,C) And a preset coefficient threshold Ass _Min Comparing;

when Ass _(A,B,C) ≥Ass _Min When the correlation is described as: the constitution to be determined has correlation among the tongue region image, the face image and the eyebrow image;

when Ass _(A,B,C) ＜Ass _Min When the correlation is described as: the constitution to be determined has no correlation among the tongue region image, the face image and the eyebrow image.

Specifically, the correlation coefficient is calculated by the following formula:

wherein A is the median of the matching values of the constitution to be determined in n times of output through the tongue picture identification model, B is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model, C is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model,for the average value of the matching values output by the tongue picture identification model for the undetermined physique several times,/I>For said constitution to be determined, an average value of the matching values outputted several times by said facial image discrimination model,/->And (3) outputting an average value of matching values for the undetermined physique through the object identification model for a plurality of times.

The application is based on the correlation coefficient Ass _(A,B,C) The influence of errors of single numerical values on the result is reduced, so that the accuracy of the final constitution identification result is ensured.

Based on the above, the judging conditions are: and identifying whether the obtained physique has correlation in the tongue region image, the face image and the eyebrow image.

Further, when the correlation description of the tongue region identification value, the face image identification value and the eye image identification value does not meet a preset judgment condition, acquiring one or more of the tongue region image, the face image and the eye image again, acquiring the corresponding tongue region identification value, the face image identification value or the eye image identification value, and then performing the correlation description. And when the repeated correlation description does not meet the preset judging condition, respectively inputting each obtained tongue area identification value, the facial image identification value or the physique type corresponding to the object identification value into the medicated diet collocation model to recommend medicated diet materials.

The medicated diet material can be recommended by any one of the prior art, for example: based on the matching of the data (constitution type and preset medicated diet materials). However, in order to further ensure the rationality of the medicated diet material and comprehensively consider the pharmacological characteristics of the combination of multiple medicinal materials, as a preferred implementation manner of this embodiment, the method for identifying and matching the medicated diet scheme based on deep learning of physique further includes:

classifying the recommended medicated diet materials according to the corresponding physique types, and grading the favorability of the medicated diet materials on the physique according to the pharmacology of the traditional Chinese medicinal materials;

sequentially arranging the obtained phenotype physique and the obtained stealth physique according to the tongue region identification value, the facial image identification value and the object image identification value to obtain a physique sequence Seq _CON ，Seq _CON = { constitution 1, constitution 2,., constitution N };

calculate the medicated diet value CP, cp=k ₁ *M _m1.1 +K ₂ *M _m2.1 +...+K _N *M _mN.1 Wherein M is _mN.1 A medicinal material of a plurality of medicinal materials corresponding to constitution N is shown, K is a medication decision coefficient, and K ₁ ,K ₂ ...K _N E (0, 1), when K _i When the medicine is=0, the corresponding medicine and other medicines are in pharmacological conflict, otherwise, the corresponding medicine is not in pharmacological conflict with other medicines;

and calculating a medicated diet value CP corresponding to all medicinal material combinations in the medicated diet materials, and selecting a group of most optimal medicated diet materials with the largest medicated diet value CP.

By means of the above, through calculation of the medicated diet value CP, not only can medicated diet materials which are more beneficial to the corresponding physique be recommended, but also medical history conflicts possibly occurring among the medicinal materials can be avoided, so that the use safety of users is ensured, and the technical feasibility and reliability are improved.

Further, in order to reduce the difficulty in selecting a user, in this embodiment, when a medicinal diet CP is calculated, and a pharmacological conflict exists between a medicinal material corresponding to the appearance type physique and other medicinal materials corresponding to the stealth type physique, K corresponding to the appearance type physique in the medicinal diet CP is adjusted to 1, and K corresponding to the stealth type physique in the medicinal diet CP is adjusted to 0. The beneficial effect of setting like this is, can be abundant give the recommendation result of multiple medicated diet material that corresponds to different physique types, provide more medicated diet material selections for the user, reduce user's selection degree of difficulty.

In summary, according to the method for identifying and matching the medicated diet scheme based on the deep learning, the constitution is identified through the deep learning of various characterization data, and the constitution identification result of the user is obtained, so that the constitution type of the user is known, and then the matching of the medicated diet materials is performed based on the constitution type, so that the common user can clearly know the medicated diet suitable for the constitution of the common user, and great convenience is provided for the user. In addition, the accuracy of the constitution identification result is ensured in the data processing process of constitution identification, and the recommending process of the medicated diet (material) is strict, so that the reliability and the practicability of the technology are ensured.

In a second aspect, the present embodiment discloses a system for identifying and matching a medicated diet scheme based on deep learning, which is applicable to the method for identifying and matching a medicated diet scheme based on deep learning. Specifically, the system for identifying and matching the medicated diet scheme based on the constitution of deep learning comprises:

the data acquisition module is configured to: an authentication parameter for collecting a user, the authentication parameter comprising: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

the model training module comprises a tongue image identification model unit, a facial image identification model unit, a eye image identification model unit and a medicated diet collocation model unit; the tongue image identification model unit is configured to: based on deep learning, training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes to obtain a tongue image identification model; the face image discrimination model unit is configured to: training by taking facial features and corresponding facial colors as feature attributes based on deep learning to obtain a facial image identification model; the object discrimination model unit is configured to: training by taking an eyebrow image and corresponding eye socket color, eyeball color and eyeball shape as characteristic attributes based on deep learning to obtain an eye image identification model; the medicated diet collocation model unit is configured as follows: based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model; wherein the tongue image identification model is configured to: based on an input tongue region image, carrying out feature analysis on the tongue region image to obtain a tongue image feature attribute corresponding to the tongue region image, and calculating a matching value of the obtained tongue image feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue region identification value; the facial image authentication model is configured to: performing feature analysis on the face image based on the input face image to obtain a corresponding face feature attribute, and performing matching value calculation on the obtained face feature attribute and a preset physique result based on a convolutional neural network to obtain a face identification value; the object identification model is configured to: based on an input eyebrow image, performing feature analysis on the eyebrow image to obtain corresponding object feature attributes, and calculating a matching value of the obtained object feature attributes and a preset physique result based on a convolutional neural network to obtain an object identification value;

a correlation evaluation module configured to: performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, comparing the obtained correlation description with preset judging conditions, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value as a phenotype constitution when the correlation description meets the preset judging conditions, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value as a stealth constitution;

the medicated diet collocation model is configured as follows: and recommending the medicated diet materials by taking the phenotype physique and the stealth physique as input layers and based on the types of the phenotype physique and the stealth physique.

It should be noted that, by means of the above-mentioned system for identifying and matching a medicated diet scheme based on deep learning according to the present embodiment, the technical effects of the system for identifying and matching a medicated diet scheme based on deep learning disclosed in the present embodiment are the same as the above-mentioned method for identifying and matching a medicated diet scheme based on deep learning, and the technical means not disclosed in the system for identifying and matching a medicated diet scheme based on deep learning are not described herein, and the relevant descriptions in the above-mentioned method for identifying and matching a medicated diet scheme based on deep learning can be referred to correspondingly, which is not described herein.

In this embodiment, for a software implementation, some or all of the flow of the embodiments may be accomplished by a computer program to instruct the associated hardware. When implemented, the above-described programs may be stored in or transmitted as one or more instructions or code on a computer-readable storage medium. In particular, computer-readable storage 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 may be any available media that can be accessed by a computer. The computer-readable storage media may include, but is not limited to, 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.

Finally, it should be noted that: the foregoing description is only a preferred embodiment of the present application, and although the present application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the technical solutions described in the foregoing embodiments, or equivalents may be substituted for some of the technical features thereof, and any modifications, equivalents, improvements or changes that fall within the spirit and principles of the present application are intended to be included in the scope of protection of the present application.

Claims (10)

1. A method for identifying and matching a medicated diet scheme based on deep learning physique, which is characterized by comprising the following steps:

training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes based on deep learning to obtain a tongue image identification model, training by taking facial features and corresponding face colors as characteristic attributes to obtain a face image identification model, and training by taking eyebrow images and corresponding eye socket colors, eyeball colors and eyeball forms as characteristic attributes to obtain an eye image identification model;

based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model;

inputting the tongue image identification model, the face image identification model and the object image identification model corresponding to identification parameters of a user, wherein the identification parameters comprise: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

the tongue picture identification model performs feature analysis on the tongue picture image based on the input tongue picture image to obtain a tongue picture feature attribute corresponding to the tongue picture image, and performs matching value calculation on the obtained tongue picture feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue picture identification value; the face image identification model performs feature analysis on the face image based on the input face image to obtain corresponding face image feature attributes, and performs matching value calculation on the obtained face image feature attributes and a preset physique result based on a convolutional neural network to obtain a face image identification value; the eye image identification model performs feature analysis on the eye image based on the input eye image to obtain corresponding eye image feature attributes, and performs matching value calculation on the obtained eye image feature attributes and a preset physique result based on a convolutional neural network to obtain an eye image identification value;

performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a phenotype constitution, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value to be used as a hidden constitution when the correlation description of the tongue region identification value, the facial image identification value and the object identification value meets a preset judgment condition;

inputting the phenotype physique and the invisible physique into the medicated diet collocation model, wherein the medicated diet collocation model recommends medicated diet materials based on the phenotype physique and the type of physique in the invisible physique.

2. The method for deep learning based constitution identification and matching medicated diet regimen of claim 1, wherein the obtaining of identification parameters further comprises: and carrying out color averaging and normalization processing on the tongue region image, the face image and the eyebrow image.

3. The method for identifying and matching medicated diet scheme based on deep learning according to claim 1, wherein the correlation description specifically comprises:

defining a constitution of the same type corresponding to the tongue region discrimination value, the facial image discrimination value and the eye image discrimination value as a constitution to be determined, and calculating a correlation coefficient Ass of the constitution to be determined in the tongue region image, the facial image and the eye-brow image _(A,B,C) ；

The obtained correlation coefficient Ass _(A,B,C) And a preset coefficient threshold Ass _Min Comparing;

when Ass _(A,B,C) ≥Ass _Min When the correlation is described as: the constitution to be determined has correlation among the tongue region image, the face image and the eyebrow image;

when Ass _(A,B,C) ＜Ass _Min When the correlation is described as: the constitution to be determined has no correlation among the tongue region image, the face image and the eyebrow image.

4. The method for deep learning based constitution identification and matching medicated diet regimen according to claim 3, wherein the correlation coefficient is calculated by the following formula:

wherein A is the median of the matching values of the constitution to be determined in n times of output through the tongue picture identification model, B is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model, C is the median of the matching values of the constitution to be determined in n times of output through the face picture identification model,for the average value of the matching values output by the tongue picture identification model for the undetermined physique several times,/I>For said constitution to be determined, an average value of the matching values outputted several times by said facial image discrimination model,/->And (3) outputting an average value of matching values for the undetermined physique through the object identification model for a plurality of times.

5. The method for identifying and matching medicated diet according to claim 3, wherein the judging condition is: and identifying whether the obtained physique has correlation in the tongue region image, the face image and the eyebrow image.

6. The method for deep learning based constitution discrimination and matching medicated diet scheme according to claim 1, wherein when a correlation description of said tongue region discrimination value, said face image discrimination value and said eye image discrimination value does not satisfy a preset judgment condition, one or more of said tongue region image, said face image and said eye image is re-acquired, and then the corresponding tongue region discrimination value, said face image discrimination value or said eye image discrimination value is obtained, and then the correlation description is performed.

7. The method for deep learning based constitution identification and matching medicated diet scheme according to claim 6, wherein when the repeated correlation description does not meet the preset judgment condition, each obtained tongue area identification value, the facial image identification value or the constitution type corresponding to the target image identification value is respectively input into the medicated diet collocation model to recommend medicated diet materials.

8. The method for identifying and matching a medical diet regimen based on deep learning according to claim 1, wherein the method for identifying and matching a medical diet regimen based on deep learning further comprises:

classifying the recommended medicated diet materials according to the corresponding physique types, and grading the favorability of the medicated diet materials on the physique according to the pharmacology of the traditional Chinese medicinal materials;

sequentially arranging the obtained phenotype physique and the obtained stealth physique according to the tongue region identification value, the facial image identification value and the object image identification value to obtain a physique sequence Seq _CON ，Seq _CON = { constitution 1, constitution 2,., constitution N };

calculate the medicated diet value CP, cp=k ₁ *M _m1.1 +K ₂ *M _m2.1 +...+K _N *M _mN.1 Wherein M is _mN.1 A medicinal material of a plurality of medicinal materials corresponding to constitution N is shown, K is a medication decision coefficient, and K ₁ ,K ₂ ...K _N E (0, 1), when K _i When the medicine is=0, the corresponding medicine and other medicines are in pharmacological conflict, otherwise, the corresponding medicine is not in pharmacological conflict with other medicines;

and calculating a medicated diet value CP corresponding to all medicinal material combinations in the medicated diet materials, and selecting a group of most optimal medicated diet materials with the largest medicated diet value CP.

9. The method for identifying and matching medicated diet scheme based on deep learning according to claim 8, wherein when the medicated diet value CP is calculated, when pharmacological conflict exists between the medicinal material corresponding to the phenotype physique and the medicinal material corresponding to other stealth physique, K corresponding to the phenotype physique in the medicated diet value CP is adjusted to 1, and K corresponding to the stealth physique in the medicated diet value CP is adjusted to 0.

10. A deep learning-based constitution identification and matching medicated diet scheme system, which is suitable for the deep learning-based constitution identification and matching medicated diet scheme method according to any one of claims 1 to 9, and is characterized in that the deep learning-based constitution identification and matching medicated diet scheme system comprises:

the data acquisition module is configured to: an authentication parameter for collecting a user, the authentication parameter comprising: a tongue region image acquired in a non-invasive way, a face image obtained by shooting the face of a user and an eyebrow image obtained by shooting the eyebrow of the user are adopted;

the model training module comprises a tongue image identification model unit, a facial image identification model unit, a eye image identification model unit and a medicated diet collocation model unit; the tongue image identification model unit is configured to: based on deep learning, training by taking tongue fur images and corresponding tongue colors, fur colors and tongue images as characteristic attributes to obtain a tongue image identification model; the face image discrimination model unit is configured to: training by taking facial features and corresponding facial colors as feature attributes based on deep learning to obtain a facial image identification model; the object discrimination model unit is configured to: training by taking an eyebrow image and corresponding eye socket color, eyeball color and eyeball shape as characteristic attributes based on deep learning to obtain an eye image identification model; the medicated diet collocation model unit is configured as follows: based on deep learning, training by taking different types of physique types and corresponding medicinal diet materials as characteristic attributes to obtain a medicinal diet collocation model; wherein the tongue image identification model is configured to: based on an input tongue region image, carrying out feature analysis on the tongue region image to obtain a tongue image feature attribute corresponding to the tongue region image, and calculating a matching value of the obtained tongue image feature attribute and a preset physique result based on a convolutional neural network to obtain a tongue region identification value; the facial image authentication model is configured to: performing feature analysis on the face image based on the input face image to obtain a corresponding face feature attribute, and performing matching value calculation on the obtained face feature attribute and a preset physique result based on a convolutional neural network to obtain a face identification value; the object identification model is configured to: based on an input eyebrow image, performing feature analysis on the eyebrow image to obtain corresponding object feature attributes, and calculating a matching value of the obtained object feature attributes and a preset physique result based on a convolutional neural network to obtain an object identification value;

a correlation evaluation module configured to: performing correlation description on the tongue region identification value, the facial image identification value and the object identification value, comparing the obtained correlation description with preset judging conditions, extracting the constitution identification results of the same type corresponding to the tongue region identification value, the facial image identification value and the object identification value as a phenotype constitution when the correlation description meets the preset judging conditions, and collecting the constitution identification results of different types corresponding to the tongue region identification value, the facial image identification value and the object identification value as a stealth constitution;

the medicated diet collocation model is configured as follows: and recommending the medicated diet materials by taking the phenotype physique and the stealth physique as input layers and based on the types of the phenotype physique and the stealth physique.