CN116503841A - A mental health intelligent emotion recognition method - Google Patents

Abstract

The invention provides a mental health intelligent emotion recognition method, which comprises the following steps: a user terminal and a service cloud; the service cloud creates a model, trains the model, adjusts model parameters and identifies the subsequent service after emotion; the user terminal is responsible for shooting, storing and expression recognition of voice and pictures and uploading the voice and pictures to the service remote end. And the server side continuously trains a model according to the collected characteristic expression data set, optimizes model parameters, forms an effective model algorithm and continuously improves the accuracy and the effectiveness of emotion recognition. The face photo of the driver can be acquired in real time, the facial expression is recognized, and a preliminary emotion recognition report is provided.

Description

A mental health intelligent emotion recognition method

technical field

The invention relates to the field of emotion recognition, in particular to a mental health intelligent emotion recognition method.

Background technique

Psychological emotion is a way of people's feedback to objective things, and it is also a way of presenting people's emotions. As an important part of human psychological activities, emotions play a pivotal role in organizing and guiding people's behavior, communicating and predicting the wishes of others. Due to factors such as physiological changes, facial expression changes, and voice changes, people express their emotions and generate corresponding signals, and cause others to infer. This process is the process of emotion recognition.

Good mood is an Internet medical platform focusing on mental and mental health services in the CNS (central nervous system). Through a large amount of data and literature analysis, it discusses the definition and classification of emotions, and discusses the future direction of emotion recognition research and its application value. outlook.

For a long time, the accident rate of long-distance truck drivers has been relatively high, resulting in relatively large personal and property losses. Through follow-up analysis, most of the accidents were caused by improper driving operations caused by truck drivers’ large emotional fluctuations. Insurance In order to provide better services and reduce the risk of accidents, the company proposes to monitor truck drivers’ facial expressions in real time, analyze truck drivers’ emotions, and report to the server in time when the driver has a dangerous emotion. The driver's emotions can regulate the driver's safe driving behavior, thereby reducing the probability of accidents.

Based on the analysis of the above problems, the product solutions currently on the market are synthesized. The proposed product solution is to create a good face recognition model by entering and modeling the face of the truck driver. The terminal device detects the facial feature photos of the driving position in real time, identifies the identity of the truck driver, and recognizes the facial expressions of the truck driver while driving through the facial expression photos, and analyzes the truck driver's emotions comprehensively in the time dimension. When the truck driver's emotions appear Abnormal, report analysis results in real time, photo information to the server, and generate an alarm. The server receives the alarm information, manually analyzes the dangerous behavior, sends out the command through the server, sends out the voice package, and the product service mode of warm voice reminder behavior, effectively terminates the driver's dangerous driving emotions and behaviors, and achieves the service mode of reducing accidents .

Contents of the invention

In view of the above problems, the present invention is proposed to provide a mental health intelligent emotion recognition method that overcomes the above problems or at least partially solves the above problems.

According to one aspect of the present invention, a mental health intelligent emotion recognition method is provided, and the emotion recognition method includes: a user terminal and a service cloud;

The service cloud creates a model, model training, model parameter adjustment and follow-up services after emotion recognition;

The user terminal is responsible for taking, storing, and facial expression recognition of voice and photos, and uploading them to the remote service end.

Optionally, the method for emotion recognition specifically includes:

Upload the face photo to the service cloud through the user terminal;

The service cloud passes the face photo through the face analysis model, creates face recognition model parameters, and then sends the parameters to the model of the user terminal for face recognition of the terminal;

The user terminal regularly takes photos, recognizes faces, analyzes the results of facial expressions, analyzes the current emotions of people, and uploads the analysis results to the service cloud.

Optionally, after uploading the analysis results to the service cloud, it also includes:

The service cloud provides follow-up services differently according to different emotional results.

Optionally, the face recognition algorithm specifically includes: a face recognition algorithm, and a sentiment analysis algorithm that dynamically adjusts the algorithm model and algorithm parameters as business volume increases, business scenarios change, and data volume increases.

Optionally, the expression recognition result will be deeply integrated with the business line to provide auxiliary support for the business line's decision-making.

A mental health intelligent emotion recognition method provided by the present invention, the emotion recognition method includes: a user terminal and a service cloud; the service cloud creates a model, model training, model parameter adjustment and follow-up services after emotion recognition; the user The terminal is responsible for taking, storing, and facial expression recognition of voice and photos, and uploading them to the remote end of the service. Based on the collected characteristic expression data sets, the server continuously trains the model, optimizes the model parameters, forms an effective model algorithm, and continuously improves the accuracy and effectiveness of emotion recognition. It can collect the driver's face photos in real time, recognize facial expressions, and issue a preliminary emotion recognition report.

The above description is only an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the contents of the description, and in order to make the above and other purposes, features and advantages of the present invention more obvious and understandable , the specific embodiments of the present invention are enumerated below.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For Those of ordinary skill in the art can also obtain other drawings based on these drawings without making creative efforts.

Fig. 1 is the flowchart of facial emotion recognition provided by the embodiment of the present invention;

Fig. 2 is a flowchart of a method for facial emotion recognition + voice emotion recognition provided by an embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided for more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

The terms "comprising" and "having" and any variations thereof in the description, embodiments, claims and drawings of the present invention are intended to cover non-exclusive inclusion, for example, including a series of steps or units.

The technical solutions of the present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments.

Create a good face recognition model by entering and modeling the face of the truck driver. The terminal device detects the facial feature photos of the driving position in real time, identifies the identity of the truck driver, and recognizes the facial expressions of the truck driver while driving through the facial expression photos, and analyzes the truck driver's emotions comprehensively in the time dimension. When the truck driver's emotions appear Abnormal, report analysis results in real time, photo information to the server, and generate an alarm. The server receives the alarm information, manually analyzes the dangerous behavior, sends out the command through the server, sends out the voice package, and the product service mode of warm voice reminder behavior, effectively terminates the driver's dangerous driving emotions and behaviors, and achieves the service mode of reducing accidents .

The current product solution is to remotely input face information on the server side, generate basic face modeling information, and deliver the face model to the terminal device. The terminal device is embedded with face recognition algorithms and expression recognition algorithms, and the terminal device controls the photos. Input frequency, frequency of expression recognition, save key feature expression data, and upload to the server. The terminal device can receive voice playback instructions from the server, and algorithm parameter adjustment and algorithm upgrade from the server.

Based on the collected characteristic expression data sets, the server continuously trains the model, optimizes the model parameters, forms an effective model algorithm, and continuously improves the accuracy and effectiveness of emotion recognition. Provide more, better and higher-quality follow-up services by continuously enriching the content and presentation of the voice service package.

At present, it is possible to collect the driver's face photos in real time, recognize facial expressions, and issue a preliminary emotion recognition report.

Emotion recognition includes two parts: user terminal and server cloud. The cloud is responsible for model creation, model training, model parameter adjustment, and subsequent service provision after emotion recognition. The terminal is responsible for voice and photo shooting, storage, expression recognition, uploading to the cloud and other functions.

The basic technical route of emotion recognition is: upload face photos to the cloud through the terminal, and the cloud passes the photos through the face analysis model to create face recognition model parameters, and then send the parameters to the terminal model for face recognition on the terminal. The terminal takes photo groups at regular intervals, recognizes faces, analyzes the results of facial expressions, analyzes the current emotions of people, and uploads the analysis results to the cloud. According to different emotional results, the cloud provides follow-up services differently.

The face recognition algorithm and emotion analysis algorithm will dynamically adjust the algorithm model and algorithm parameters as the business volume increases, the business scene changes, and the data volume increases. Continuously improve the accuracy of the algorithm and the business scenario coverage of the algorithm.

Expression recognition is currently only implanted in the Linux system, and will be compatible with other terminal devices in the future, covering applications in multiple scenarios. With the deepening of the algorithm, there will be a greater demand for hardware computing power in the future, and it is bound to be deeply bound with hardware manufacturers to promote the development of hardware.

The results of expression recognition will be deeply integrated with other business lines to provide auxiliary support for the decision-making of other business lines.

The core technology of face emotion recognition + speech emotion recognition is divided into two levels. One is the algorithm model, and the other is the compatibility of multi-scenario devices.

In terms of algorithm models, the current face recognition algorithms and expression recognition algorithms on the market have the characteristics of low recognition degree and simple recognition results, especially for the voice emotion recognition data set, the algorithm recognition degree is particularly low. Our algorithm will continuously improve the recognition accuracy of the algorithm model, the accuracy of expression recognition, the multi-scenario coverage of emotion analysis, and the expression change trend of a single person based on business scenarios and a large amount of data training. Build a model of the basic characteristics of facial expressions of people in different regions and occupations, and model the emotional characteristics of patients with depression.

Currently, face recognition + expression recognition algorithm models are mostly set on fixed terminals. For various mobile terminals, industrial terminals have little coverage. Most of the hardware devices on the market do not support the implantation of algorithm models, especially the dynamic upgrade of algorithm models is not operable. Industrial-grade hardware is also expensive and has limited computing power. Our company can continuously cover business scenarios and reduce costs according to business scenarios, so that this technology can benefit more people.

At present, the application of face algorithm at the industrial level is basically implanted in the basic C language, but a large number of algorithm models are in python language.

As shown in Figure 1, facial emotion recognition is based on artificial intelligence technology and is used to analyze emotions from different pictures and videos. Generally, signals are obtained from cameras, social media pages, video libraries, etc., static and dynamic facial expression detection is performed, and then emotional states are classified; facial emotion recognition is usually based on deep learning algorithms, after face information preprocessing, feature learning, emotion There are three stages of identification; in the field of mental illness, facial emotion recognition technology can be used to predict the probability of illness, assist in diagnosis, and assist in improving the quality of care for medical personnel. At present, facial emotion recognition has inherent risks such as data accuracy, algorithm fairness, data privacy, and user reactivity.

As shown in Figure 2, speech emotion recognition, that is, given a speech signal, the computer automatically judges the multi-dimensional information of the speaker. Human speech production is a complex multi-system coordination process involving brain cognitive activities and body muscle movements. Speech signals contain three layers of information: acoustics, language, and emotion. The motor fibers are highly mechanized, and the generated speech signals have objective and repeatable characteristics. The speech recognition process goes through three stages: speech signal processing, feature extraction, and emotion modeling. The recognition algorithms involved include traditional algorithms, algorithms based on deep learning, and end-to-end algorithms. Speech biomarkers are features or feature combinations of speech and audio signals related to clinical outcomes, which can be used for screening, diagnosis, and disease detection of mental and psychological diseases, as well as intervention methods for AI+CBT and digital therapy. At present, due to the vague definition of emotion, the scarcity of data and the difficulty of labeling, speech emotion recognition technology is difficult.

Facial emotion recognition + voice emotion recognition can effectively improve the recognition accuracy, and can mutually verify the data interference factors generated in the recognition process, and the user's emotions affect the interference factors and affect the accuracy of the results.

Beneficial effects: the server remotely enters face information, generates basic face modeling information, and sends the face model to the terminal device. The terminal device is embedded with a face recognition algorithm, an expression recognition algorithm, and the terminal device controls the frequency of photo entry, expression The frequency of recognition, save key feature expression data, and upload to the server. The terminal device can receive voice playback instructions from the server, and algorithm parameter adjustment and algorithm upgrade from the server.

Based on the collected characteristic expression data sets, the server continuously trains the model, optimizes the model parameters, forms an effective model algorithm, and continuously improves the accuracy and effectiveness of emotion recognition. Provide more, better and higher-quality follow-up services by continuously enriching the content and presentation of the voice service package.

At present, it has been able to collect the driver's face photos in real time, recognize facial expressions, and issue a preliminary emotion recognition report.

The above specific implementation manners have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation modes of the present invention, and are not used to limit the protection scope of the present invention. Within the spirit and principles of the present invention, any modifications, equivalent replacements, improvements, etc., shall be included in the protection scope of the present invention.

Claims (5)

1. A mental health intelligent emotion recognition method is characterized in that, the emotion recognition method comprises: a user terminal and a service cloud; The service cloud creates a model, model training, model parameter adjustment and follow-up services after emotion recognition; The user terminal is responsible for taking, storing, and facial expression recognition of voice and photos, and uploading them to the remote service end. 2. a kind of mental health intelligent emotion recognition method according to claim 1, is characterized in that, the method for described emotion recognition specifically comprises: Upload the face photo to the service cloud through the user terminal; The service cloud passes the face photo through the face analysis model, creates face recognition model parameters, and then sends the parameters to the model of the user terminal for face recognition of the terminal; The user terminal regularly takes photos, recognizes faces, analyzes the results of facial expressions, analyzes the current emotions of people, and uploads the analysis results to the service cloud. 3. A kind of mental health intelligent emotion recognition method according to claim 2, is characterized in that, after described analysis result is uploaded to described service cloud, also comprises: The service cloud provides follow-up services differently according to different emotional results. 4. A mental health intelligent emotion recognition method according to claim 2, characterized in that, the face recognition algorithm specifically comprises: a face recognition algorithm, and the emotion analysis algorithm will increase with the increase in business volume, Changes, increase in the amount of data, and dynamically adjust the algorithm model and algorithm parameters. 5. A mental health intelligent emotion recognition method according to claim 1, characterized in that the result of the facial expression recognition will be deeply integrated with the business line to provide auxiliary support for the decision-making of the business line.