Facial feature recognition method and device based on traditional Chinese medicine knowledge, equipment and storage medium
By combining multi-spectral and visible light data models, the facial feature recognition model is constructed, which solves the problem of scene diversity and insufficient data of facial feature recognition methods in traditional Chinese medicine knowledge, improves the accuracy and efficiency of feature recognition, and promotes the development of traditional Chinese medicine facial diagnosis technology .
Patent Information
- Application Number
- CN202411185453.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-06
AI Technical Summary
The facial feature recognition method based on traditional Chinese medicine knowledge in the prior art is difficult to deal with various scenarios, the feature recognition accuracy and efficiency are poor, and the data acquisition and labeling cost is high, and the generalization ability is limited.
Using facial feature recognition method based on traditional Chinese medicine knowledge, the target feature recognition model is constructed by obtaining the pending facial images and the Chinese medicine knowledge labeled image set, combining the multi-spectral data model (Xception model) and the visible light data model (MobileNetV2 model), and the target feature recognition model is constructed, and the Bay Leaf is used The optimization algorithm combines feature vectors to perform feature extraction and classification to improve model generalization capabilities.
It improves the efficiency and accuracy of facial feature recognition, fills the gap in facial feature recognition in traditional Chinese medicine knowledge, improves the intelligence level of traditional Chinese medicine facial diagnosis, and provides a more objective, accurate and comprehensive reference basis for traditional Chinese medicine diagnosis.
Smart Images

Figure CN119091490A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a facial feature recognition method, device, equipment and storage medium based on traditional Chinese medicine knowledge. Background Art
[0002] In the existing technology applications, deep learning technology has been widely used in the field of medical image analysis. By building and training neural network models, it can automatically extract features from images, and perform classification and prediction. However, when using traditional Chinese medicine knowledge for facial feature recognition, the application of deep learning technology is still in its early stages. Due to the diversity and complexity of data, existing deep learning models may have insufficient generalization capabilities in practical applications and are difficult to cope with various complex scenarios. At the same time, the training of deep learning models requires a large amount of high-quality annotated data, while the annotated data using traditional Chinese medicine knowledge is relatively small, and the cost of data acquisition and annotation is high, resulting in insufficient data for model training and failure to meet model performance requirements. Summary of the invention
[0003] The main purpose of this application is to provide a facial feature recognition method, device, equipment and storage medium based on traditional Chinese medicine knowledge, aiming to solve the technical problems that the facial feature recognition method based on traditional Chinese medicine knowledge in the prior art is difficult to cope with various scenarios and the accuracy and efficiency of the feature recognition method are poor.
[0004] To achieve the above objectives, the present application proposes a facial feature recognition method based on TCM knowledge, the facial feature recognition method based on TCM knowledge comprising:
[0005] Obtaining facial images to be processed and a set of images annotated with traditional Chinese medicine knowledge;
[0006] A model is constructed according to the TCM knowledge annotated image set, the multispectral data model and the visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model;
[0007] Inputting the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed;
[0008] A feature recognition result of the facial image to be processed is obtained according to the integrity feature information and the gloss feature information.
[0009] In one embodiment, the step of constructing a model based on the TCM knowledge annotated image set, the multispectral data model, and the visible light data model to obtain a target feature recognition model further includes:
[0010] Perform feature extraction based on the multispectral annotation data set and the multispectral data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of first feature vectors;
[0011] Perform feature extraction based on the visible light annotation data set and the visible light data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of second feature vectors;
[0012] Determine a plurality of fused feature vectors according to each first feature vector, each second feature vector and a target Bayesian optimization algorithm;
[0013] The target classifier is trained according to each fused feature vector to obtain a target feature recognition model.
[0014] In one embodiment, the step of extracting features based on the multispectral annotation data set and the multispectral data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of first feature vectors includes:
[0015] Determine a plurality of multispectral annotated images according to the multispectral annotated data set in the traditional Chinese medicine knowledge annotated image set;
[0016] Perform data preprocessing on each multispectral annotated image to obtain multiple processed facial images;
[0017] Each processed facial image is input into a multispectral data model for feature extraction to obtain a plurality of first feature vectors.
[0018] In one embodiment, the step of determining a plurality of fused feature vectors according to each first feature vector, each second feature vector and a target Bayesian optimization algorithm comprises:
[0019] Dynamically adjusting the weighting coefficients of the multispectral data model and the visible light data model according to a target Bayesian optimization algorithm to obtain a target weighting coefficient;
[0020] Each first eigenvector and each second eigenvector are fused according to the target weighting coefficient to obtain a plurality of fused eigenvectors.
[0021] In one embodiment, the step of training the target classifier according to each fused feature vector to obtain the target feature recognition model includes:
[0022] The target classifier is trained according to each fused feature vector to obtain a first training model;
[0023] Aggregating the multi-spectral annotation dataset and the visible light annotation dataset in the traditional Chinese medicine knowledge annotation image set to obtain a sample training image set;
[0024] The first training model is trained according to the sample training image set to obtain a target feature recognition model.
[0025] In one embodiment, the step of training the first training model according to the sample training image set to obtain a target feature recognition model includes:
[0026] Fine-tune the first training model according to the sample training image set and the target learning rate scheduler to obtain a second training model;
[0027] Performing a performance evaluation on the second training model to obtain a performance evaluation result;
[0028] When the performance evaluation result is a preset qualified result, the second training model is format converted according to the TensorFlow lightweight format to obtain a target feature recognition model.
[0029] In one embodiment, before the step of obtaining the facial image to be processed and the image set annotated with TCM knowledge, the method further includes:
[0030] Obtain a visible light facial image set and a multispectral annotated image set;
[0031] Performing TCM knowledge annotation on the visible light facial image set and the multi-spectral annotated image set respectively to obtain a visible light annotated data set and a multi-spectral annotated data set;
[0032] A traditional Chinese medicine knowledge annotated image set is generated according to the visible light annotated data set and the multi-spectrum annotated data set.
[0033] In addition, to achieve the above purpose, the present application also proposes a facial feature recognition device based on TCM knowledge, the facial feature recognition device based on TCM knowledge comprising:
[0034] An acquisition module, used to acquire the facial image to be processed and the image set annotated with traditional Chinese medicine knowledge;
[0035] A construction module is used to construct a model according to the TCM knowledge annotated image set, a multispectral data model and a visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model;
[0036] An input module, used for inputting the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed;
[0037] A processing module is used to obtain a feature recognition result of the facial image to be processed according to the integrity feature information and the gloss feature information.
[0038] In addition, to achieve the above-mentioned purpose, the present application also proposes a facial feature recognition device based on traditional Chinese medicine knowledge, the device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the facial feature recognition method based on traditional Chinese medicine knowledge as described above.
[0039] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the facial feature recognition method based on traditional Chinese medicine knowledge as described above are implemented.
[0040] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the facial feature recognition method based on traditional Chinese medicine knowledge as described above.
[0041] The present application obtains a facial image to be processed and a set of images annotated with TCM knowledge; constructs a model based on the set of images annotated with TCM knowledge, a multispectral data model, and a visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; inputs the facial image to be processed into the target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; and obtains a feature recognition result of the facial image to be processed based on the integrity feature information and gloss feature information. Through the above method, a target feature recognition model is constructed based on the TCM knowledge annotated image set, the multispectral data model and the visible light data model. The integrity feature information and gloss feature information of the facial image to be processed are obtained through the target feature recognition model, so as to obtain the corresponding feature recognition results. The multispectral information and TCM knowledge are fully utilized for feature recognition. While improving the efficiency and accuracy of feature recognition, it also fills the gap in facial feature recognition using TCM knowledge, and improves the generalization ability of the model in practical applications. It can be applied to various scenarios, improves the intelligence level of TCM facial diagnosis, and provides a more objective, accurate and comprehensive reference basis for TCM diagnosis, which promotes the development and application of TCM facial diagnosis technology. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0044] Figure 1 A flowchart diagram of the first embodiment of the facial feature recognition method based on traditional Chinese medicine knowledge provided in this application;
[0045] Figure 2 A flowchart diagram of the second embodiment of the facial feature recognition method based on traditional Chinese medicine knowledge provided in this application;
[0046] Figure 3 A schematic diagram of a simplified process of a facial feature recognition method based on traditional Chinese medicine knowledge provided in Example 2 of the present application;
[0047] Figure 4 This is a schematic diagram of the module structure of a facial feature recognition device based on traditional Chinese medicine knowledge in an embodiment of the present application;
[0048] Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the facial feature recognition method based on traditional Chinese medicine knowledge in the embodiment of the present application.
[0049] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0050] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.
[0051] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.
[0052] The main solution of the embodiment of the present application is: obtaining a facial image to be processed and a set of images annotated with TCM knowledge;
[0053] A model is constructed according to the TCM knowledge annotated image set, the multispectral data model and the visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; the facial image to be processed is input into the target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; and a feature recognition result of the facial image to be processed is obtained according to the integrity feature information and the gloss feature information.
[0054] At present, TCM facial diagnosis is an important part of TCM diagnosis. It judges the patient's constitution, condition and health status by observing the color, shape, texture and other characteristics of the patient's face. However, traditional TCM facial diagnosis mainly relies on the doctor's experience and subjective judgment, which has certain limitations and shortcomings. In recent years, with the rapid development of artificial intelligence and multispectral imaging technology, more and more research and application attempts have been made to introduce these technologies into TCM facial diagnosis, in order to realize the intelligence and objectivity of the facial diagnosis process.
[0055] In the existing technical applications, deep learning technology has been widely used in the field of medical image analysis. By building and training neural network models, it can automatically extract features from images, and perform classification and prediction. However, in the field of TCM facial diagnosis, the application of deep learning technology is still in its early stages. Due to the diversity and complexity of data, existing deep learning models may have insufficient generalization capabilities in practical applications and are difficult to cope with various complex facial diagnosis scenarios. At the same time, the training of deep learning models requires a large amount of high-quality annotated data, while the annotated data in the field of TCM facial diagnosis is relatively small, and the cost of data acquisition and annotation is high.
[0056] For spectral detection, existing research focuses on using single spectral images or limited bands for facial feature recognition and analysis, while multispectral imaging technology can obtain image information of objects in multiple bands. Compared with single spectral imaging, it can provide more spectral features and detailed information, thereby improving the accuracy and reliability of image analysis. In addition, current research focuses on multispectral imaging systems for facial diagnosis, and there are few methods for judging TCM facial symptoms based on the features obtained by multispectral imaging technology, which need further exploration.
[0057] At present, the existing TCM facial diagnosis technologies and products have the following key problems: 1. Strong data dependence: The training of deep learning models requires a large amount of high-quality labeled data, while the labeled data in the field of TCM facial diagnosis is relatively small, and the cost of data acquisition and labeling is high. 2. Limited model generalization ability: Due to the diversity and complexity of data, existing deep learning models may have insufficient generalization capabilities in practical applications and are difficult to cope with different facial diagnosis scenarios. 3. Insufficient use of spectral information: Current research focuses on multispectral imaging systems for facial diagnosis, and there are few methods for TCM facial diagnosis symptom judgment based on the features obtained by multispectral imaging technology, which need further exploration. The above-mentioned TCM facial diagnosis is to obtain the user's facial features and lay the foundation for subsequent disease diagnosis, not to directly diagnose the disease.
[0058] The present application provides a solution, which constructs a target feature recognition model based on a set of images annotated with TCM knowledge, a multispectral data model, and a visible light data model, and obtains the integrity feature information and gloss feature information of the facial image to be processed through the target feature recognition model, thereby obtaining the corresponding feature recognition results, and making full use of multispectral information and TCM knowledge for feature recognition. While improving the efficiency and accuracy of feature recognition, it also fills the gap in facial feature recognition using TCM knowledge, and improves the generalization ability of the model in practical applications. It can be applied to various scenarios, improves the intelligence level of TCM facial diagnosis, provides a more objective, accurate and comprehensive reference basis for TCM diagnosis, and promotes the development and application of TCM facial diagnosis technology.
[0059] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a facial feature recognition device based on traditional Chinese medicine knowledge that can realize the above functions. The following takes the facial feature recognition device based on traditional Chinese medicine knowledge as an example of the execution subject to illustrate this embodiment and the following embodiments.
[0060] Based on this, the present application embodiment provides a facial feature recognition method based on traditional Chinese medicine knowledge, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the facial feature recognition method based on traditional Chinese medicine knowledge of the present application.
[0061] In this embodiment, the facial feature recognition method based on traditional Chinese medicine knowledge includes steps S10 to S40:
[0062] Step S10, obtaining the facial image to be processed and the image set annotated with TCM knowledge.
[0063] It should be noted that the facial image to be processed is an image that needs to detect and classify multiple features in the face, including but not limited to facial gloss, facial integrity and other facial features. The TCM knowledge image set includes but is not limited to a visible light annotation dataset and a multispectral annotation dataset. In this embodiment, the visible light annotation dataset includes a large number of visible light facial images annotated with facial features by TCM physicians or TCM artificial intelligence devices; the multispectral annotation dataset includes a large number of multispectral facial images annotated with facial features by TCM physicians or TCM artificial intelligence devices. In this embodiment, in order to facilitate the understanding of facial feature labels, facial features are illustrated by examples through Table 1, and other features and corresponding label contents may also be included, which are not limited in this embodiment.
[0064] Table 1
[0065] Facial features Tagged content Facial Gloss Glossy, matte Facial integrity Complete, incomplete
[0066] In a feasible implementation manner, before step S10, steps A11 to A13 may be included:
[0067] Step A11, obtaining a visible light facial image set and a multi-spectral annotated image set.
[0068] It should be noted that, in this embodiment, a large number of multispectral annotated images are obtained through a multispectral imaging bio-information acquisition device, thereby forming a multispectral annotated image set. In this embodiment, the multispectral imaging bio-information acquisition device is mainly composed of an electrically tunable co-filter, a high-gain camera, and a portable spectrometer. The filter and the high-gain camera are used to acquire multispectral images, and the spectrometer is used to acquire the spectral curve of the experimental object. The computer controls the electrically tunable co-filter to select the band and spectral spacing of 400 to 720 nm (non-visible light range), and simultaneously controls the parameters, display, and image acquisition of the high-gain camera.
[0069] It can be understood that, in this embodiment, a large number of visible light facial images can be collected by an industrial-grade camera in an environment where light can be controlled, thereby forming a visible light facial image set.
[0070] Step A12, respectively annotating the visible light facial image set and the multi-spectral annotated image set with TCM knowledge to obtain a visible light annotated data set and a multi-spectral annotated data set.
[0071] It should be noted that facial features are annotated on each visible light facial image in the visible light facial image set by a Chinese medicine physician or a Chinese medicine artificial intelligence device, and facial features are annotated on each multispectral annotated image in the multispectral annotated image set by a Chinese medicine physician or a Chinese medicine artificial intelligence device. All annotated visible light facial images constitute a visible light annotated data set, and all annotated multispectral annotated images constitute a multispectral annotated data set.
[0072] It can be understood that in this embodiment, in order to facilitate the classification and annotation of facial images, an online data annotation platform can be developed, and the annotation personnel such as Chinese medicine doctors or Chinese medicine artificial intelligence equipment only need to enter the online data annotation platform and select the corresponding parameters for each facial image to complete the annotation.
[0073] Step A13, generating a TCM knowledge annotated image set according to the visible light annotated dataset and the multi-spectrum annotated dataset.
[0074] It should be noted that the TCM knowledge annotated image set is composed of a visible light annotated dataset and a multispectral annotated dataset. In the present embodiment, the visible light annotated dataset and the multispectral annotated dataset are stored separately in the TCM knowledge annotated image set, so as to facilitate the rapid acquisition of visible light annotated images or multispectral annotated images during subsequent model construction.
[0075] Step S20, constructing a model based on the TCM knowledge annotated image set, the multispectral data model and the visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model.
[0076] It should be noted that when building the model, the GPU is configured to allow for on-demand growth, and the required libraries and modules are imported, including TensorFlow, Pandas, Matplotlib, etc. This ensures the normal operation of the deep learning environment and improves computing efficiency.
[0077] It is understandable that in this embodiment, the annotated images in the TCM knowledge annotated image set are preprocessed to ensure that the image data format and quality meet the model requirements, and the Xception model is used to process the multispectral annotated data in the TCM knowledge annotated image, and the MobileNetV2 model is used to process the visible light annotated data in the TCM knowledge annotated image. The weighted average fusion model with the weighted coefficient adjusted by the Bayesian optimization algorithm is further used as the basic model, and the multi-task face feature recognition model is constructed in combination with data enhancement, global average pooling and attention mechanism processing. Again, the established model is trained and fine-tuned at a lower learning rate, and the model parameters are updated by the back propagation algorithm to optimize the performance of the model on the training set and the validation set. Then, the changes in the loss function, precision and recall rate during the training process are analyzed, and the model weight with the lowest loss in the validation set is selected as the optimal model, and saved. Finally, the trained optimal model is converted to the TensorFlowLite format to improve the reasoning efficiency and real-time performance on the mobile device, and the model after the format conversion is the target feature recognition model.
[0078] In this embodiment, the Xception model is a deep learning model whose core idea is to separate the layer-by-layer convolution and channel-by-channel convolution (also called depth-separable convolution) in the traditional Inception module, which can reduce the number of parameters and improve the efficiency of the model; the MobileNetV2 model is an efficient convolutional neural network model that aims to maintain high accuracy while reducing computational complexity and memory usage.
[0079] Step S30, inputting the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed.
[0080] It should be noted that the facial image to be processed is input into the target feature recognition model, and the target feature recognition model can output the gloss feature information and integrity feature information of the facial image to be processed. In this embodiment, the gloss feature information can reflect whether the user's face in the facial image to be processed is glossy, and the integrity feature information can reflect whether the user's face in the facial image to be processed is complete.
[0081] Step S40, obtaining a feature recognition result of the facial image to be processed according to the integrity feature information and the gloss feature information.
[0082] It should be noted that the feature recognition results of the facial image to be processed can be obtained through the integrity feature information and gloss feature information output by the target feature recognition model, providing a more objective, accurate and comprehensive reference for subsequent TCM diagnosis.
[0083] This embodiment obtains a facial image to be processed and a set of images annotated with TCM knowledge; constructs a model based on the TCM knowledge annotated image set, a multispectral data model, and a visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; inputs the facial image to be processed into the target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; and obtains a feature recognition result of the facial image to be processed based on the integrity feature information and the gloss feature information. Through the above method, a target feature recognition model is constructed based on the TCM knowledge annotated image set, the multispectral data model and the visible light data model. The integrity feature information and gloss feature information of the facial image to be processed are obtained through the target feature recognition model, so as to obtain the corresponding feature recognition results. The multispectral information and TCM knowledge are fully utilized for feature recognition. While improving the efficiency and accuracy of feature recognition, it also fills the gap in facial feature recognition using TCM knowledge, and improves the generalization ability of the model in practical applications. It can be applied to various scenarios, improves the intelligence level of TCM facial diagnosis, and provides a more objective, accurate and comprehensive reference basis for TCM diagnosis, which promotes the development and application of TCM facial diagnosis technology.
[0084] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 3 In the facial feature recognition method based on traditional Chinese medicine knowledge, step S20 further includes steps S21 to S24:
[0085] Step S21, performing feature extraction based on the multispectral annotation data set and the multispectral data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of first feature vectors.
[0086] It should be noted that, in this embodiment, the multispectral data model is composed of an Xception model, and each multispectral annotated image in the multispectral annotated data set is preprocessed and then input into the multispectral data model for feature extraction, thereby obtaining a feature vector corresponding to each multispectral annotated image. In this embodiment, the feature vector corresponding to a multispectral annotated image is a first feature vector.
[0087] In a feasible implementation, step S21 may include steps B11 to B13:
[0088] Step B11, determining a plurality of multispectral annotated images according to the multispectral annotated data set in the traditional Chinese medicine knowledge annotated image set.
[0089] Step B12, performing data preprocessing on each multi-spectral annotated image to obtain a plurality of processed facial images.
[0090] It should be noted that, in the present embodiment, the CSV file formed by the TCM knowledge annotation image set is used to extract the annotated data, generate a label dictionary, save it in a pickle file, and create an image path list and corresponding labels. At the same time, decoding, scaling and normalization preprocessing operations are performed. Data loading and preprocessing modules are directly connected to the data set (including annotated facial image data), and the raw data is converted into a format that the model can process, and the processed image data is passed to the next stage.
[0091] It is understandable that data enhancement operations such as random cropping, brightness adjustment, contrast adjustment, and random flipping are performed on each multispectral annotated image to improve the generalization ability of the model. The data enhancement module works in parallel with the multispectral data model, and the enhanced image is passed to the normalization layer for processing. In this embodiment, the multispectral annotated image after data preprocessing is the processed facial image.
[0092] In the specific implementation, the data set is divided into training set, validation set and test set, and cache and batch processing optimization are performed to ensure the reasonable division of the data set and improve the efficiency of data loading and processing.
[0093] Step B13: input each processed facial image into a multispectral data model for feature extraction to obtain a plurality of first feature vectors.
[0094] It should be noted that the multispectral data model is used to extract high-level features in each processed facial image, thereby obtaining a first feature vector corresponding to each processed facial image. In this embodiment, the multispectral data model receives preprocessed image data as input, extracts features through a convolutional layer, adds data enhancement, a global average pooling layer and an attention mechanism, and constructs an output layer.
[0095] Step S22, performing feature extraction based on the visible light annotation data set and the visible light data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of second feature vectors.
[0096] It should be noted that, in this embodiment, the visible light data model is composed of a MobileNetV2 model, and each visible light annotated image in the visible light annotated data set is preprocessed and then input into the visible light data model for feature extraction, thereby obtaining a feature vector corresponding to each visible light annotated image. In this embodiment, a feature vector corresponding to a visible light annotated image is a second feature vector.
[0097] It can be understood that the visible light data model and the multispectral data model share the same input layer, and the same data enhancement technology is applied to the input visible light annotated image. Features are extracted through the visible light data model, and the features are mapped to a vector of fixed length after global average pooling, and the attention mechanism is introduced for processing. In this embodiment, by adding a model suitable for visible light data, the overall recognition ability can be enhanced.
[0098] Step S23, determining a plurality of fused feature vectors according to each first feature vector, each second feature vector and a target Bayesian optimization algorithm.
[0099] It should be noted that the outputs of the visible light data model and the multi-spectral data model are weighted averaged by the target Bayesian optimization algorithm, so as to obtain a plurality of fused feature vectors obtained by fusing each first feature vector and each second feature vector.
[0100] In a feasible implementation manner, step S23 may include steps C11 to C12:
[0101] Step C11, dynamically adjusting the weighting coefficients of the multispectral data model and the weighting coefficients of the visible light data model according to a target Bayesian optimization algorithm to obtain target weighting coefficients.
[0102] Step C12: performing vector fusion on each first eigenvector and each second eigenvector according to the target weighting coefficient to obtain a plurality of fused eigenvectors.
[0103] It should be noted that the weighting coefficients of the multispectral data model and the visible light data model are dynamically adjusted by the target Bayesian optimization algorithm to obtain the optimization weights corresponding to the two models. In this embodiment, the target weighting coefficients include the optimization weights of the multispectral data model and the optimization weights of the visible light data model.
[0104] It is understandable that each first eigenvector and each second eigenvector are weighted averaged according to the target weight coefficient to obtain multiple fused eigenvectors. In this embodiment, the Bayesian optimization algorithm is used to dynamically adjust the weight coefficients of the Xception and MobileNetV2 models, and the outputs of the two models are weighted averaged according to the optimized weights. As the objective function, the average accuracy of the model on the validation set is used to evaluate the performance of the weighted average model. Multiple parallel classifiers process the output eigenvectors respectively and perform multi-task classification, including facial gloss and facial integrity.
[0105] Step S24, training the target classifier according to each fused feature vector to obtain a target feature recognition model.
[0106] It should be noted that the target classifier is trained according to multiple fused feature vectors, and then the learning rate scheduler is introduced to fine-tune the model. Finally, the trained optimal model is converted into the TensorFlow Lite format to obtain a target feature recognition model that can perform facial feature recognition.
[0107] In a feasible implementation, step S24 may include steps D11 to D13:
[0108] Step D11, training the target classifier according to each fused feature vector to obtain a first training model.
[0109] It should be noted that the target classifier is trained based on multiple fused feature vectors, and the trained target classifier is the first training model. In this embodiment, the target classifier receives multiple fused feature vectors and inputs the probability distribution of each feature category to obtain the final classification result. Through the above training, the overall recognition accuracy of the model can be improved.
[0110] Step D12, aggregating the multi-spectral annotation data set and the visible light annotation data set in the traditional Chinese medicine knowledge annotated image set to obtain a sample training image set.
[0111] It should be noted that, since the multispectral annotation dataset and the visible light annotation dataset in the TCM knowledge annotation image set are stored separately, in order to further improve the performance of the model, each multispectral annotation image and each visible light annotation image are mixed without distinguishing between visible light images and multispectral images, thereby forming a sample training image set. In this embodiment, the sample training image set contains a large number of facial annotation images.
[0112] Step D13: training the first training model according to the sample training image set to obtain a target feature recognition model.
[0113] It should be noted that a learning rate scheduler is introduced, the first training model is fine-tuned through a sample training image set and the learning rate scheduler, and the trained optimal model is converted into the TensorFlow Lite format, thereby obtaining a target feature recognition model that can perform facial feature recognition.
[0114] In a feasible implementation manner, step D13 may include steps E11 to E13:
[0115] Step E11, fine-tuning the first training model according to the sample training image set and the target learning rate scheduler to obtain a second training model.
[0116] It should be noted that the first training model is trained using the sample training image set, the basic fusion model is unfrozen, the target learning rate scheduler is invaded, and fine-tuning is performed at a lower learning rate to optimize the accuracy and generalization ability of the target classifier. Connected to the target classifier, the model parameters are updated through the back propagation algorithm, and the optimal weights are saved according to the validation set performance. In this embodiment, the first training model that has been trained is the second training model.
[0117] Step E12: performing performance evaluation on the second training model to obtain a performance evaluation result.
[0118] Step E13, when the performance evaluation result is a preset qualified result, the second training model is format converted according to the TensorFlow lightweight format to obtain a target feature recognition model.
[0119] It should be noted that merging the training history and drawing the loss and accuracy curves can intuitively display the training process of the model, and introduce evaluation indicators such as recall rate to comprehensively evaluate the performance of the second training model. When the performance evaluation result of the second training model is a preset qualified result, it means that the deep learning model that can be used for facial feature recognition has been trained. At this time, the second training model is converted to TensorFlow Lite format (i.e., TensorFlow lightweight format), and the second training model after format conversion is the target feature recognition model. In this embodiment, the second training model with qualified performance evaluation results will be saved, and a lightweight target feature recognition file will be output for deployment.
[0120] This embodiment obtains multiple first feature vectors by performing feature extraction based on the multispectral annotation data set and the multispectral data model in the TCM knowledge annotated image set; obtains multiple second feature vectors by performing feature extraction based on the visible light annotation data set and the visible light data model in the TCM knowledge annotated image set; determines multiple fused feature vectors based on each first feature vector, each second feature vector and the target Bayesian optimization algorithm; trains the target classifier based on each fused feature vector to obtain a target feature recognition model. In the above manner, the performance and lightweight of the model are guaranteed.
[0121] For example, in order to help understand the implementation process of the facial feature recognition method based on TCM knowledge obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 3 , Figure 3 A brief schematic diagram of a facial feature recognition method based on TCM knowledge is provided, specifically:
[0122] The method of this embodiment uses visible light + multi-spectral face images, combined with different deep learning models (Xception + MobileNetV2 models), and introduces a variety of data enhancement techniques, attention mechanisms and learning rate scheduling strategies to achieve multi-task classification of face images, improve the accuracy and robustness of the model, and is suitable for a variety of practical application scenarios. The system is built using the TensorFlow and Keras frameworks, and uses TensorFlow Lite for model optimization so that it can be quickly deployed and run on devices with limited resources.
[0123] Through the solution of this embodiment, multiple features in facial images, such as facial gloss and facial integrity, can be detected and classified simultaneously. It can solve the problems of low accuracy in facial feature recognition, poor generalization ability, and insufficient use of spectral information in the prior art, and can be applied in fields such as traditional Chinese medicine medical diagnosis to improve the application effect and universality of artificial intelligence in image recognition.
[0124] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the facial feature recognition method based on traditional Chinese medicine knowledge of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.
[0125] This application also provides a facial feature recognition device based on traditional Chinese medicine knowledge, please refer to Figure 4 , the facial feature recognition device based on traditional Chinese medicine knowledge comprises:
[0126] The acquisition module 10 is used to acquire the facial image to be processed and the image set annotated with TCM knowledge.
[0127] The construction module 20 is used to construct a model based on the TCM knowledge annotated image set, the multispectral data model and the visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model.
[0128] The input module 30 is used to input the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed.
[0129] The processing module 40 is used to obtain a feature recognition result of the facial image to be processed according to the integrity feature information and the gloss feature information.
[0130] Optionally, the building module 20 is further used to:
[0131] Feature extraction is performed based on the multi-spectral annotation data set and the multi-spectral data model in the TCM knowledge annotated image set to obtain multiple first feature vectors; feature extraction is performed based on the visible light annotation data set and the visible light data model in the TCM knowledge annotated image set to obtain multiple second feature vectors; multiple fused feature vectors are determined based on each first feature vector, each second feature vector and a target Bayesian optimization algorithm; a target classifier is trained based on each fused feature vector to obtain a target feature recognition model.
[0132] Optionally, the building module 20 is further used to:
[0133] Determine a plurality of multispectral annotated images according to the multispectral annotated data set in the traditional Chinese medicine knowledge annotated image set; perform data preprocessing on each multispectral annotated image to obtain a plurality of processed facial images; input each processed facial image into a multispectral data model for feature extraction to obtain a plurality of first feature vectors.
[0134] Optionally, the building module 20 is further used to:
[0135] According to the target Bayesian optimization algorithm, the weighting coefficients of the multispectral data model and the visible light data model are dynamically adjusted to obtain the target weighting coefficients; according to the target weighting coefficients, each first eigenvector and each second eigenvector are vector fused to obtain multiple fused eigenvectors.
[0136] Optionally, the building module 20 is further used to:
[0137] The target classifier is trained according to each fused feature vector to obtain a first training model; the multi-spectral annotation data set and the visible light annotation data set in the traditional Chinese medicine knowledge annotated image set are aggregated to obtain a sample training image set; the first training model is trained according to the sample training image set to obtain a target feature recognition model.
[0138] Optionally, the building module 20 is further used to:
[0139] The first training model is fine-tuned according to the sample training image set and the target learning rate scheduler to obtain a second training model; a performance evaluation is performed on the second training model to obtain a performance evaluation result; when the performance evaluation result is a preset qualified result, the second training model is format converted according to the TensorFlow lightweight format to obtain a target feature recognition model.
[0140] Optionally, the acquisition module 10 is further used to:
[0141] Obtain a visible light facial image set and a multispectral annotated image set; perform TCM knowledge annotation on the visible light facial image set and the multispectral annotated image set respectively to obtain a visible light annotated data set and a multispectral annotated data set; generate a TCM knowledge annotated image set based on the visible light annotated data set and the multispectral annotated data set.
[0142] The facial feature recognition device based on TCM knowledge provided by the present application adopts the facial feature recognition method based on TCM knowledge in the above-mentioned embodiment, which can solve the technical problems that the facial feature recognition method based on TCM knowledge in the prior art is difficult to cope with various scenarios, and the accuracy and efficiency of the feature recognition method are poor. Compared with the prior art, the beneficial effects of the facial feature recognition device based on TCM knowledge provided by the present application are the same as the beneficial effects of the facial feature recognition method based on TCM knowledge provided by the above-mentioned embodiment, and the other technical features of the facial feature recognition device based on TCM knowledge are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.
[0143] The present application provides a facial feature recognition device based on traditional Chinese medicine knowledge, and the facial feature recognition device based on traditional Chinese medicine knowledge includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the facial feature recognition method based on traditional Chinese medicine knowledge in the above-mentioned embodiment one.
[0144] Reference below Figure 5, which shows a schematic diagram of the structure of a facial feature recognition device based on TCM knowledge suitable for implementing the embodiment of the present application. The facial feature recognition device based on TCM knowledge in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The facial feature recognition device based on traditional Chinese medicine knowledge shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
[0145] like Figure 5 As shown, the facial feature recognition device based on TCM knowledge may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the facial feature recognition device based on TCM knowledge are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the facial feature recognition device based on TCM knowledge to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a facial feature recognition device based on TCM knowledge with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have alternatively.
[0146] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.
[0147] The facial feature recognition device based on TCM knowledge provided by the present application adopts the facial feature recognition method based on TCM knowledge in the above-mentioned embodiment, which can solve the technical problems that the facial feature recognition method based on TCM knowledge in the prior art is difficult to cope with various scenarios, and the accuracy and efficiency of the feature recognition method are poor. Compared with the prior art, the beneficial effects of the facial feature recognition device based on TCM knowledge provided by the present application are the same as the beneficial effects of the facial feature recognition method based on TCM knowledge provided by the above-mentioned embodiment, and the other technical features of the facial feature recognition device based on TCM knowledge are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.
[0148] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.
[0149] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0150] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, the computer-readable program instructions being used to execute the facial feature recognition method based on traditional Chinese medicine knowledge in the above-mentioned embodiment.
[0151] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.
[0152] The computer-readable storage medium may be included in the facial feature recognition device based on traditional Chinese medicine knowledge; or may exist independently without being assembled into the facial feature recognition device based on traditional Chinese medicine knowledge.
[0153] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by a facial feature recognition device based on TCM knowledge, the facial feature recognition device based on TCM knowledge: obtains a facial image to be processed and a set of images annotated with TCM knowledge; constructs a model based on the set of images annotated with TCM knowledge, a multispectral data model, and a visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; inputs the facial image to be processed into the target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; and obtains a feature recognition result of the facial image to be processed based on the integrity feature information and gloss feature information.
[0154] Computer program code for performing the operations of the present application may be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0155] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0156] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.
[0157] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned facial feature recognition method based on TCM knowledge, and can solve the technical problems that the facial feature recognition method based on TCM knowledge in the prior art is difficult to cope with various scenarios, and the accuracy and efficiency of the feature recognition method are poor. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the facial feature recognition method based on TCM knowledge provided by the above-mentioned embodiment, and will not be repeated here.
[0158] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned facial feature recognition method based on traditional Chinese medicine knowledge.
[0159] The computer program product provided by the present application can solve the technical problems that the facial feature recognition method based on TCM knowledge in the prior art is difficult to cope with various scenarios, and the accuracy and efficiency of the feature recognition method are poor. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the facial feature recognition method based on TCM knowledge provided by the above embodiment, and will not be repeated here.
[0160] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.
Claims
1. A facial feature recognition method based on traditional Chinese medicine knowledge, characterized in that: The facial feature recognition method based on traditional Chinese medicine knowledge comprises: Obtaining facial images to be processed and a set of images annotated with traditional Chinese medicine knowledge; A model is constructed according to the TCM knowledge annotated image set, the multispectral data model and the visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; Inputting the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; A feature recognition result of the facial image to be processed is obtained according to the integrity feature information and the gloss feature information.
2. The method according to claim 1, characterized in that The step of constructing a model based on the TCM knowledge annotated image set, the multispectral data model, and the visible light data model to obtain a target feature recognition model also includes: Perform feature extraction based on the multispectral annotation data set and the multispectral data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of first feature vectors; Perform feature extraction based on the visible light annotation data set and the visible light data model in the traditional Chinese medicine knowledge annotated image set to obtain a plurality of second feature vectors; Determine a plurality of fused feature vectors according to each first feature vector, each second feature vector and a target Bayesian optimization algorithm; The target classifier is trained according to each fused feature vector to obtain a target feature recognition model.
3. The method according to claim 2, characterized in that The step of extracting features based on the multispectral annotation data set and the multispectral data model in the TCM knowledge annotated image set to obtain a plurality of first feature vectors comprises: Determine a plurality of multispectral annotated images according to the multispectral annotated data set in the traditional Chinese medicine knowledge annotated image set; Perform data preprocessing on each multispectral annotated image to obtain multiple processed facial images; Each processed facial image is input into a multispectral data model for feature extraction to obtain a plurality of first feature vectors.
4. The method according to claim 2, characterized in that The step of determining a plurality of fused feature vectors according to each first feature vector, each second feature vector and a target Bayesian optimization algorithm comprises: Dynamically adjusting the weighting coefficients of the multispectral data model and the visible light data model according to a target Bayesian optimization algorithm to obtain a target weighting coefficient; Each first eigenvector and each second eigenvector are fused according to the target weighting coefficient to obtain a plurality of fused eigenvectors.
5. The method according to claim 2, characterized in that The step of training the target classifier according to each fused feature vector to obtain the target feature recognition model includes: The target classifier is trained according to each fused feature vector to obtain a first training model; Aggregating the multi-spectral annotation dataset and the visible light annotation dataset in the traditional Chinese medicine knowledge annotation image set to obtain a sample training image set; The first training model is trained according to the sample training image set to obtain a target feature recognition model.
6. The method according to claim 5, characterized in that The step of training the first training model according to the sample training image set to obtain a target feature recognition model comprises: Fine-tune the first training model according to the sample training image set and the target learning rate scheduler to obtain a second training model; Performing a performance evaluation on the second training model to obtain a performance evaluation result; When the performance evaluation result is a preset qualified result, the second training model is format converted according to the TensorFlow lightweight format to obtain a target feature recognition model.
7. The method according to any one of claims 1 to 6, characterized in that Before the step of obtaining the facial image to be processed and the image set annotated with TCM knowledge, the method further includes: Obtain a visible light facial image set and a multispectral annotated image set; Performing TCM knowledge annotation on the visible light facial image set and the multi-spectral annotated image set respectively to obtain a visible light annotated data set and a multi-spectral annotated data set; A traditional Chinese medicine knowledge annotated image set is generated according to the visible light annotated data set and the multi-spectrum annotated data set.
8. A facial feature recognition device based on traditional Chinese medicine knowledge, characterized in that: The facial feature recognition device based on traditional Chinese medicine knowledge comprises: An acquisition module, used to acquire the facial image to be processed and the image set annotated with traditional Chinese medicine knowledge; A construction module is used to construct a model according to the TCM knowledge annotated image set, a multispectral data model and a visible light data model to obtain a target feature recognition model, wherein the multispectral data model is composed of an Xception model, and the visible light data model is composed of a MobileNetV2 model; An input module, used for inputting the facial image to be processed into a target feature recognition model to obtain integrity feature information and gloss feature information of the facial image to be processed; A processing module is used to obtain a feature recognition result of the facial image to be processed according to the integrity feature information and the gloss feature information.
9. A facial feature recognition device based on traditional Chinese medicine knowledge, characterized in that: The facial feature recognition device based on TCM knowledge includes: a memory, a processor, and a facial feature recognition program based on TCM knowledge stored in the memory and executable on the processor, wherein the facial feature recognition program based on TCM knowledge is configured to implement the facial feature recognition method based on TCM knowledge as described in any one of claims 1 to 7.
10. A storage medium, characterized in that: The storage medium stores a facial feature recognition program based on TCM knowledge, and when the facial feature recognition program based on TCM knowledge is executed by the processor, the facial feature recognition method based on TCM knowledge as described in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Face gloss analytical method based on inspection diagnosis of traditional Chinese medical science
CN103034874A
Facial information analysis method and system and storage medium
CN118379780A
