An emotion recognition system and method

Abstract

This invention relates to an emotion recognition system and method. The system includes: a terminal module operated by an individual at a designated location for at least information uploading and / or publishing; a data acquisition module configured at the designated location for acquiring at least one type of information within the designated location; and a processing module communicatively connected to the terminal module and the data acquisition module. The processing module is capable of acquiring, directly and / or indirectly, feature data information containing at least image-like characteristics from the terminal module and the data acquisition module, and determining a corresponding emotion score by identifying the emotion category of facial expressions in one or more images of the feature data information. The method includes acquiring, in at least two ways, feature data information containing image-like characteristics for one or more individuals at a designated location; and determining a corresponding emotion score by identifying the emotion category of facial expressions in one or more images of the feature data information.

Description

Technical Field

[0001] This invention relates to the field of emotion recognition and processing technology, and in particular to an emotion recognition system and method. Background Technology

[0002] Numerous studies in modern medicine have shown that a patient's psychological state (emotions / feelings) often affects the progress of their treatment and the speed of recovery, sometimes even directly leading to improvement or deterioration of their condition. Furthermore, a patient's psychological state is closely related to their quality of life. Many clinical cases also reflect that disease development and psychological changes are closely related, even mutually causal. For example, 30% of patients visiting the gastroenterology department also have depressive or anxiety disorders. Depressive and anxiety disorders are among the most common mental disorders. Besides emotional problems, they can also manifest as various physical symptoms, with the digestive system being most susceptible. Common gastrointestinal symptoms include nausea, acid reflux, constipation, heartburn, indigestion, and bloating. The prevalence of depressive disorders in people with gastrointestinal symptoms is five times higher than in the general population, and the prevalence of anxiety disorders is four times higher. Moreover, the prevalence is unrelated to the cause; the more gastrointestinal symptoms a person has, the higher their prevalence. For gastroenterology patients, especially those with a history of mental illness, anxiety, fear, or unstable mental states can significantly reduce treatment adherence and negatively impact treatment outcomes, potentially leading to doctor-patient disputes. Utilizing existing technologies to monitor patient emotions allows for the rapid determination of personalized emotional support strategies. Existing technologies for patient emotion monitoring include, for example:

[0003] CN114883014A discloses a biometric-based patient emotion feedback device, method, and treatment bed, mainly comprising the following modules: a main camera for collecting facial feature information and facial dynamic change information of the patient; a limb camera for collecting the patient's limb movements; a distributed server for analyzing the patient's real-time emotions based on the facial feature information and facial dynamic change information collected by the main camera and the limb movements collected by the limb camera, and obtaining emotion analysis results; and a warning device for issuing prompt information based on the emotion analysis results from the distributed server.

[0004] However, existing technologies for monitoring the emotions of individual patients have significant limitations. For example, in places where patients gather in large numbers, it is difficult to conduct timely and effective tracking and monitoring of each individual patient. This may lead to a large amount of negative emotions in the treatment area that are difficult to manage. In particular, people who are confined to designated places for specific reasons, such as infectious diseases, are more likely to experience negative emotions.

[0005] Infectious diseases are a class of diseases caused by various pathogens that can be transmitted between humans, animals, or between humans and animals. Most pathogens are microorganisms, while a small portion are parasites; diseases caused by parasites are also called parasitic diseases. For some infectious diseases, disease control departments must promptly monitor the incidence and take timely countermeasures. Therefore, upon discovery, they should be reported to the local disease control department within the prescribed time; these are called legally notifiable infectious diseases. For example, China has three categories of legally notifiable infectious diseases: A, B, and C, totaling 40 types.

[0006] For outbreaks of infectious diseases, especially Class A infectious diseases and some Class B infectious diseases classified as Class A by relevant national management departments, infected patients are typically confined to designated isolation facilities for centralized treatment within specified areas. They are only allowed to leave after recovery and confirmation of no further issues. Examples of such centralized isolation facilities include makeshift hospitals. Isolated patients inevitably experience significant emotional fluctuations due to their anxieties about their own health and that of their families. These negative emotions can negatively impact both the patient's recovery and the treatment efforts of medical staff; therefore, monitoring and managing their emotions is crucial. Existing technologies include:

[0007] CN215068279U provides an emotion recognition and reassurance system based on makeshift hospitals, belonging to the field of facial recognition technology. It includes a facial expression acquisition module, a data dimension converter, a feature extractor, an emotion matching module, a display module, an anxiety level matching module, and a reassurance module. The reassurance module includes multiple reassurance actuators. The facial expression acquisition module is connected to the display module via the data dimension converter, feature extractor, and emotion matching module. The input of the anxiety level matching module is connected to the output of the feature extractor, and the outputs are connected to the corresponding reassurance actuators in the reassurance module.

[0008] CN115324388A provides a psychotherapy cabin, relating to the field of medical equipment technology, including a cabin body, psychotherapy equipment, a power supply and distribution system, a communication system, and an environmental protection system. The cabin body has a door on the front wall, and the interior includes partitioned work areas and equipment rooms. The psychotherapy equipment includes an integrated psychological assessment and emotion management machine, a cathartic shouting device, and a VR psychological training device installed within the cabin. The power supply and distribution system includes a power distribution cabinet and cables. The communication system includes a ring network switch for internal communication within the cabin and for networking with external communication systems to achieve information interconnection. The environmental protection system includes a heating, ventilation, and lighting system.

[0009] Current technologies typically only utilize cameras to capture facial images of individual people, using facial recognition to determine their corresponding facial emotions, and then providing different pre-set reassurance methods to patients. However, in areas experiencing large-scale outbreaks of infectious diseases, the large number of people isolated in designated quarantine sites, their high mobility, and the complexity of their psychological states, coupled with a severe shortage or uneven distribution of medical resources (especially resources for the diagnosis and treatment of mental illnesses such as psychological intervention), make it difficult to collect the psychological state of all patients in designated quarantine sites. It also makes it difficult to provide timely and effective psychological assistance to all patients exhibiting negative emotions. Consequently, current technologies not only fail to grasp the true overall emotional state of designated quarantine sites due to the limited data collection methods and simplified facial recognition processes, but also suffer from resource waste and ineffective psychological assistance due to the irrational allocation of medical resources.

[0010] Furthermore, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making this invention, but due to space limitations, not all details and contents were listed in detail. However, this does not mean that the present invention does not possess the features of these prior art. On the contrary, the present invention already possesses all the features of the prior art, and the applicant reserves the right to add relevant prior art to the background art. Summary of the Invention

[0011] In view of the shortcomings of the prior art, the present invention provides an emotion recognition system and method to solve at least some of the above-mentioned technical problems.

[0012] This invention discloses an emotion recognition system, which includes:

[0013] The terminal module is operated by an individual at a designated location to at least enable information uploading and / or publishing; the acquisition module is configured at the designated location to acquire at least one type of information at the designated location; and the processing module is communicatively connected to the terminal module and the acquisition module.

[0014] Preferably, the processing module can obtain feature data information containing at least image properties from the terminal module and the acquisition module directly and / or indirectly, and determine the corresponding emotion score by identifying the emotion category of human expressions in one or more images of the feature data information, wherein the feature data information containing image properties is accompanied by labels with geographical and temporal attributes.

[0015] According to a preferred embodiment, when the processing module analyzes and processes the feature data information of the image properties, it can classify the collected images and / or corresponding regions in the images based at least on a first classification rule and / or a second classification rule, so as to assign corresponding processing weights.

[0016] According to a preferred embodiment, the processing module can classify images based on their source when executing the first classification rule. In this case, feature data information of the image nature uploaded and / or published by the terminal module is classified as actively captured images, and feature data information of the image nature acquired by the acquisition module is classified as passively captured images.

[0017] Preferably, the processing module can assign a higher first processing weight to passively captured images compared to actively captured images. This is because when the image-like feature data information is acquired by the shooting unit, the individual being photographed by the shooting unit usually perceives the presence of the shooting unit and is prone to producing a specific on-camera expression. This on-camera expression can be an expression displayed by the individual being photographed to show any specified emotion that is not entirely consistent with their current mood. This expression is usually unnatural (e.g., exaggerated, restrained, or stiff) or does not fully represent the individual's current mood. The specified emotion can be self-specified or specified by others. For example, an individual in a designated quarantine area may experience a depressed mood due to infectious diseases, but when invited to be photographed, they might, out of respect or goodwill, display a self-specified happy expression that is not entirely consistent with their current mood (depressed mood) during the photograph, thus making the credibility of the emotion represented by the actively captured image lower. In other words, the emotions of people shown in actively captured images may contain some false information. When the processing module has difficulty in judging the authenticity of the expressions in the footage, it can ensure the credibility of the results by appropriately reducing its first processing weight. However, when the processing module is able to judge the authenticity of the expressions in the footage, it can increase its first processing weight.

[0018] According to a preferred embodiment, when executing the second classification rule, the processing module can classify the images according to the area where the people in the images are located, so as to divide any image into a primary image area and a secondary image area.

[0019] Preferably, for actively captured images, the processing module can assign a lower second processing weight to the primary image region compared to the secondary image region; for passively captured images, the processing module can assign a higher second processing weight to the primary image region compared to the secondary image region. This is because in actively captured images, the main person occupying the primary image region is more likely to be aware of the subject's position than the secondary person occupying the secondary image region. This awareness means that the individual is more easily aware of the subject being photographed, allowing the main person in the actively captured image to typically display facial expressions and / or adjust the feature data information of the photographing unit. Furthermore, when the main person adjusts the feature data information of the photographing unit, they usually do not adjust the presentation of the secondary person in the image. Therefore, the processing module can allocate a relatively higher second processing weight to the secondary image region of the actively captured image. In passively captured images, the acquisition module targets random selection and the acquisition distance is relatively far. The acquired images typically have a clearer primary image region than the secondary image region, which is more conducive to the processing module recognizing facial expressions. Therefore, the processing module can allocate a relatively higher second processing weight to the primary image region of the passively captured image.

[0020] According to a preferred embodiment, the processing module can perform emotion recognition on any collected feature data information containing image properties based on a trained emotion judgment model to obtain the emotion value of a single image.

[0021] According to a preferred embodiment, the processing module can obtain the sentiment statistical estimate of all or part of a specified location within any time period based on the geographical and temporal attribute labels attached to the feature data information.

[0022] According to a preferred embodiment, the terminal module integrates or is connected to a camera unit to acquire image-like feature data information with tags attached to geographic and time attributes. The terminal module is capable of uploading the feature data information to the processing module and / or publishing it to a social network that can be retrieved by the processing module.

[0023] According to a preferred embodiment, the blowing module can adjust the local information collection frequency based on the distribution of individuals within a specified location according to their disease status. The adjustment of the local information collection frequency can be achieved at least by adjusting the collection frequency of the collection units whose collected feature data information has specified geographic attribute tags, and / or by adjusting the collection frequency of the processing module for feature data information with specified geographic attribute tags in the social network.

[0024] This invention also discloses an emotion recognition method, which includes:

[0025] Acquire image-based feature data information for one or more individuals within a specified location using at least two methods;

[0026] By identifying the emotion category of facial expressions in one or more images based on feature data, a corresponding emotion score can be determined.

[0027] In calculating sentiment scores, images and / or corresponding regions within images can be assigned corresponding processing weights based on different classification rules, so as to obtain a sentiment statistical estimate of all or part of a specified location within any given time period through weight processing.

[0028] According to a preferred embodiment, the acquisition of image-like feature data information includes at least two methods: active uploading and / or publishing by individuals and acquisition by means of a collection module deployed in a designated location. The local information collection frequency of any acquisition method can be adjusted according to the distribution of individuals in the designated location based on their disease status. Attached Figure Description

[0029] Figure 1 This is a simplified schematic diagram of the module connection relationship of an emotion recognition system according to a preferred embodiment of the present invention.

[0030] List of reference numerals

[0031] 100: Terminal module; 200: Acquisition module; 300: Processing module. Detailed Implementation

[0032] The following is a detailed explanation with reference to the accompanying drawings.

[0033] Figure 1 This is a simplified schematic diagram of the module connection relationship of an emotion recognition system according to a preferred embodiment of the present invention.

[0034] Example 1

[0035] This invention discloses an emotion recognition system, which can also be used for individuals whose freedom of movement is partially restricted for legitimate reasons. These individuals are allowed to move within a designated area, but are generally not permitted to leave that area within a specified time period or without the necessary authorization. Such individuals may include those under centralized quarantine due to infectious diseases, or those under centralized management for violating laws and regulations. Preferably, the emotion recognition system of this invention is particularly suitable for individuals whose freedom of movement is partially restricted for legitimate reasons but whose freedom of communication is not restricted. For example, individuals under centralized quarantine due to infectious diseases, although unable to leave their designated location during quarantine, can connect to the network via mobile terminals (e.g., mobile phones, tablets, laptops) to send and receive information. Preferably, the term "emotion" in this invention can also refer to terms with roughly the same meaning, such as "feeling," "emotion," "mood," or "psychological state."

[0036] According to a preferred embodiment, the emotion recognition system of this embodiment can be configured for people centrally quarantined due to infectious diseases. Legally notifiable infectious diseases are generally classified into three categories: A, B, and C. Typically, designated quarantine locations can be set up in areas experiencing outbreaks of Class A infectious diseases or some Class B infectious diseases. The emotion recognition system of this embodiment can be used within these designated locations (or designated quarantine points). For example, the aforementioned Class A infectious diseases may include plague and cholera, and the aforementioned "some" Class B infectious diseases may include some Class B infectious diseases treated as Class A infectious diseases by relevant national management departments, such as SARS, highly pathogenic avian influenza, and H1N1 influenza. Preferably, in this embodiment, the people centrally quarantined due to infectious diseases can be referred to as the target population, and any independent individual within the target population can be referred to as an individual.

[0037] Preferably, the emotion recognition system may include a processing module 300 with communication capabilities. The processing module 300 can communicate with the terminal module 100 to receive data uploaded by the terminal module 100, analyze and process the received data, and obtain corresponding processing results. Preferably, the terminal module 100 can be any device capable of sending and receiving information used by an individual, such as a mobile terminal. When an individual enters a designated isolation point, they can communicate with the processing module 300 via the terminal module 100. While uploading data, they can also obtain necessary relevant information from the processing module 300, particularly the operational status of the designated isolation point, such as current environmental parameters, internal personnel configuration, and resource allocation patterns.

[0038] Preferably, the terminal module 100 can guide individuals to upload feature data information within a specified time by configuring a corresponding reward mechanism. The feature data information can be presented in at least the form of images and / or videos to represent the individual's current state. Further, the terminal module 100 typically integrates a camera unit or is connected to a separate camera unit to acquire feature data information in the form of images and / or videos. A terminal module 100 integrating a camera unit can be, for example, a mobile phone, tablet computer, or laptop computer with a camera. A terminal module 100 connected to a separate camera unit can be, for example, a mobile terminal connected to a camera / camera via wired or wireless means. This mobile terminal may or may not integrate a camera unit. Preferably, the feature data information may also include text information, wherein the terminal module 100 at least has a text input function.

[0039] Preferably, after acquiring feature data information that can characterize the current state of an individual through the shooting unit, the terminal module 100 can choose whether to upload it to the processing module 300 and / or publish it to a social network based on the wishes of its user (i.e., the individual using the terminal module 100), so that the processing module 300 can directly receive the feature data information uploaded by the terminal module 100 and / or indirectly obtain the feature data information published by the terminal module 100 through retrieval and filtering in the social network.

[0040] Preferably, the image-like feature data acquired by the shooting unit is typically accompanied by tags with corresponding geographical and temporal attributes. The geographical attributes can include latitude and longitude, altitude, etc., to determine the acquisition location of the feature data, allowing the processing module 300 to determine the affiliation of the acquisition location based on the pre-entered geographical attributes of the designated isolation point's area. The temporal attribute can be the acquisition time of the feature data. Furthermore, the processing module 300 can filter feature data whose acquisition location does not belong to the designated isolation point's area, and only receive feature data belonging to the designated isolation point's area for analysis and processing, thereby reducing the computational load on the processing module 300 and achieving efficient utilization of computing power. Preferably, when uploading and / or publishing image-like feature data, the terminal module 100 can simultaneously send its corresponding geographical and temporal attribute tags, and these tags are assigned by the shooting unit at the corresponding acquisition location and acquisition time, not by the terminal module 100 during uploading and / or publishing.

[0041] Preferably, the emotion recognition system may further include several data acquisition modules 200 distributed at various locations within a designated isolation point to acquire characteristic data information of one or more individuals from different positions and angles. The data acquisition modules 200 can be configured in a manner that does not affect the normal life of the target population and / or infringe upon the legitimate rights and interests of the individuals. Preferably, the characteristic data information acquired by the data acquisition modules 200 can be presented at least in the form of images and / or videos. Further, the characteristic data information acquired by the data acquisition modules 200 may also include voice.

[0042] Preferably, the acquisition module 200 is typically positioned at a fixed location within a designated isolation point, allowing the processing module 300 to determine its corresponding geographic attributes based on the initial setup parameters of the acquisition module 200. Furthermore, the feature data information emitted by the acquisition module 200 is at least tagged with its corresponding geographic and temporal attributes.

[0043] Preferably, for the feature data information containing video, the processing module 300 can use frame number as the basis for slicing, and select some or all slices for analysis and processing based on set rules or randomly, so that the video can be converted into one or more images. Preferably, the selected image slices ensure that at least most of the facial expressions of the people contained can be recognized. Furthermore, when the video contains audio, the processing module 300 can convert the audio into text to form feature data information containing images and text.

[0044] Preferably, for feature data information containing text or text converted from video and speech, the processing module 300 can extract keywords from the text and associate the keywords with the limiting words stored in the database. By selecting the limiting words with relatively higher relevance, the preset corresponding sentiment is determined and used as auxiliary calibration information. The limiting words stored in the database can be updated by the processing module 300 by obtaining hot words from social networks.

[0045] Preferably, for the feature data information (or image-related feature data information) containing images or images converted from video images, the processing module 300 can divide the acquired image-related feature data information into several categories based on preset classification rules. The preset classification rules may have the following settings: a first classification rule is based on the image source, and a second classification rule is based on the area where the person in the image is located. Preferably, any individual person is presented as a corresponding person image in the image.

[0046] Furthermore, according to the first classification rule, the processing module 300 can classify the image-related feature data information acquired by the shooting unit into actively captured images, and can classify the image-related feature data information acquired by the acquisition module 200 into passively captured images. That is, the processing module 300 can classify the image-related feature data information based on the shooting intention of the main person in the image.

[0047] Preferably, the processing module 300 can assign a first processing weight to the divided actively captured images and passively captured images based on the first classification rule, wherein the weight assigned by the processing module 300 can add a calibration coefficient to the credibility of the analyzed sentiment results when performing sentiment analysis.

[0048] Preferably, the processing module 300 can assign a higher first processing weight to passively captured images compared to actively captured images. This is because when the image-like feature data information is acquired by the shooting unit, the individual being photographed by the shooting unit usually perceives the presence of the shooting unit and is prone to produce a specific on-camera expression. This on-camera expression can be an expression displayed by the individual being photographed to show any specified emotion that is not entirely consistent with their current mood. This expression is usually unnatural (e.g., exaggerated, restrained, or stiff) or does not fully represent the individual's current mood. The specified emotion can be self-specified or specified by others. For example, an individual in a designated quarantine area may experience a depressed mood due to infectious diseases, but when invited to be photographed, they might, out of respect or goodwill, display a self-specified happy expression that is not entirely consistent with their current mood (depressed mood) during the photograph, thus making the credibility of the person's emotion in the actively captured image relatively lower. In other words, the emotions of people shown in actively captured images may contain some false information. When it is difficult for the processing module 300 to determine the authenticity of the expressions in the footage, it can ensure the credibility of the result by appropriately reducing its first processing weight. However, when the processing module 300 is sufficient to determine the authenticity of the expressions in the footage, it can increase its first processing weight.

[0049] The passively captured images obtained by the acquisition module 200 are often distributed in a relatively concealed or inconspicuous manner throughout the designated isolation point. Even if some acquisition modules 200 are detectable when an individual enters the designated isolation point, as the isolation period continues, the individual, whose attention has been diverted, will usually find it difficult to focus on the deployment of the acquisition modules 200, and will instead focus more on treatment and daily life. Therefore, the processing module 300 can assign a relatively higher first processing weight to the passively captured images. Specifically, for passively captured images obtained by different acquisition modules 200, the processing module 300 can at least determine the corresponding first processing weight based on the placement of the acquisition modules 200 within the designated isolation point.

[0050] Furthermore, although passively captured images usually have a higher initial processing weight than actively captured images, actively captured images are also an important data source for the processing module 300 to process data. This can not only expand the data source of the processing module 300, but also enable the processing module 300 to focus on and / or make suggestions to certain special individuals. For example, if some people whose facial images are difficult to obtain by the acquisition module 200 post characteristic data information of extremely negative emotions on social networks through the terminal module 100, the processing module 300 can only determine their true state by actively capturing images, and then provide assistance by notifying relevant personnel (such as medical staff, especially those with psychological counseling capabilities).

[0051] Furthermore, the processing module 300 can classify and delineate image regions according to the second classification rule. That is, the processing module 300 can execute the second classification rule alone or execute the second classification rule together with the first classification rule. Preferably, the target of the second classification rule executed by the processing module 300 can be the feature data information of the image nature that has been classified and categorized by the first classification rule. According to the second classification rule, the processing module 300 can delineate the main image region and the secondary image region based on the relative positional relationship of the people in the image at the time of shooting, according to the feature data information of the image nature that has been classified as actively captured image or passively captured image.

[0052] Preferably, the primary and secondary image regions can be divided based on the distribution of people in the image, wherein people located in the primary image region can be primary figures, and people located in the secondary image region can be secondary figures. Preferably, in the image, primary figures typically occupy a larger spatial area compared to secondary figures. Further, primary figures can be figures whose gaze is directed or approximately directed towards the shooting unit or acquisition module 200 at the time of shooting, or figures located in the focus area of ​​the shooting unit or acquisition module 200 at the time of shooting.

[0053] Furthermore, the processing module 300 can assign second processing weights to different regions defined in the feature data information of the same image type, and the processing module 300 can assign different second processing weights to the main image region and the secondary image region defined in the corresponding image based on the difference between actively captured images and passively captured images.

[0054] Preferably, for actively captured images, the processing module 300 can assign a lower second processing weight to the primary image region of the actively captured image compared to the secondary image region; for passively captured images, the processing module 300 can assign a higher second processing weight to the primary image region of the passively captured image compared to the secondary image region. This is because, in actively captured images, the main person occupying the primary image region is more likely to be aware of the subject's position than the secondary person occupying the secondary image region. This awareness means that the individual can more easily perceive the direction being filmed, thus allowing the main person in the actively captured image to typically display facial expressions and / or adjust the feature data information of the filming unit. Furthermore, when the main person adjusts the feature data information of the filming unit, they usually do not adjust the presentation of the secondary person in the image. Therefore, the processing module 300 can allocate a relatively higher second processing weight to the secondary image region of the actively captured image. In passively captured images, the target of the acquisition module 200 is randomly selected and the acquisition distance is relatively far. The main image area is usually clearer than the secondary image area in the acquired image, which is more conducive to the processing module 300 to recognize human expressions. Therefore, the processing module 300 can allocate a relatively higher secondary processing weight to the main image area of ​​the passively captured image.

[0055] Preferably, the processing module 300 can extract and analyze different facial expressions contained in the feature data information of any image type to obtain the emotion score corresponding to the feature data information, and can obtain the emotion score of a single image based on its corresponding first processing weight and / or second processing weight and other auxiliary calibration information. Further, the processing module 300 can determine the emotion score displayed in the feature data information at least by identifying the emotion category of the corresponding facial expression, wherein the emotion category of the facial expression can be at least divided into positive emotion, neutral emotion, and negative emotion; for example, hearty laughter belongs to positive emotion, and a sorrowful expression belongs to negative emotion. Preferably, the emotion category of facial expression can be further subdivided, which can improve the recognition accuracy based on the computing power of the processing module 300. Preferably, for different emotion categories, the processing module 300 can set corresponding emotion standard values ​​to determine the emotion score of the corresponding person by identifying the emotion category corresponding to the facial expression in the image.

[0056] Preferably, the processing module 300 can perform emotion recognition on the collected image-related feature data based on a trained emotion judgment model to obtain a single image emotion value. This single image emotion value can be the average emotion score of all identified individuals in any image. Preferably, the aforementioned average emotion score can be the quotient obtained by dividing the sum of the emotion scores of each individual in the image (determined by the emotion standard value) by the number of identified individuals.

[0057] Preferably, the processing module 300 can calculate the sentiment value of a single image from several collected images to obtain a sentiment statistical estimate for the entire specified isolation point. Specifically, when calculating the sentiment statistical estimate, the processing module 300 can weight each single image sentiment value based on a first processing weight to determine the influence of various types of image-related feature data on the sentiment statistical estimate of the entire specified isolation point. Preferably, the processing module 300 can sort the collected images according to their time attribute labels in a time sequence, and can group multiple images within a preset time period as processing groups to obtain sentiment statistical estimates for different time periods. Furthermore, the processing module 300 can save the sentiment statistical estimates for several time periods. When generating a new time period's sentiment statistical estimate, the processing module 300 can delete or transfer the sentiment statistical estimate corresponding to the earliest processing group in the time sequence to free up storage space.

[0058] Preferably, the processing module 300 can be set with an emotional threshold for assessing the emotional state of a designated isolation point, so as to initiate corresponding psychological intervention measures at least when the emotional statistical estimate exceeds the emotional threshold. The emotional threshold can be determined based on factors such as the geographical area of ​​the designated isolation point, the overall patient condition at the designated isolation point, and real-time hot topics. For example, the processing module 300 can obtain current real-time hot topics by connecting to the network to assess the psychological impact trend of these hot topics on the isolated target population within the designated isolation point. More preferably, the processing module 300 can set an emotional threshold for local areas. The processing module 300 can determine the emotional statistical estimate of local areas within the designated isolation point based on the geographical attributes attached to the image-like feature data. When the emotional statistical estimate of a local area exceeds the local emotional threshold, psychological intervention measures can be initiated only in the area where negative emotions are rampant, thereby reducing the waste of medical resources and the workload of relevant staff, and ensuring the effectiveness and efficiency of the measures.

[0059] Preferably, the processing module 300 can be connected to the data management module of the medical system to obtain the overall disease status of the current designated isolation point from the data management module, thereby determining at least the distribution of individuals with different disease statuses within the target population at the designated isolation point. Further, the processing module 300 can adjust the local information collection frequency based on the distribution of individuals within the designated isolation point according to their disease status. This local information collection frequency can be achieved by adjusting the sampling frequency of the corresponding location acquisition module 200 and / or adjusting the collection frequency of feature data information with corresponding geographical attribute tags. Preferably, since the composition of the target population in the designated isolation point is constantly adjusted with the entry and exit of individuals, the overall disease status of the designated isolation point will also change in real time. Therefore, the processing module 300 can adjust the local information collection frequency in a timely manner based on the real-time changes in the overall disease status, and can also adjust the computing power allocation method so that more computing power can be allocated to calculating the sentiment statistical estimation of key local areas.

[0060] Example 2

[0061] This embodiment is a further improvement on embodiment 1, and repeated content will not be described again.

[0062] This invention also discloses an emotion recognition method, which includes:

[0063] Acquire image-based feature data information for one or more individuals within a specified location using at least two methods;

[0064] By identifying the emotion category of facial expressions in one or more images based on feature data, a corresponding emotion score can be determined.

[0065] In calculating sentiment scores, images and / or corresponding regions within images can be assigned corresponding processing weights based on different classification rules, so as to obtain a sentiment statistical estimate of all or part of a specified location within any given time period through weight processing.

[0066] According to a preferred embodiment, the acquisition of image-like feature data information includes at least the following methods: active uploading and / or publishing by individuals and acquisition by means of a collection module 200 deployed at a designated location. The local information collection frequency of any acquisition method can be adjusted according to the distribution of individuals at the designated location based on their disease status.

[0067] It should be noted that the specific embodiments described above are exemplary. Those skilled in the art can devise various solutions inspired by the disclosure of this invention, and these solutions all fall within the scope of this invention and its protection. Those skilled in the art should understand that this specification and its accompanying drawings are illustrative and do not constitute a limitation on the claims. The scope of protection of this invention is defined by the claims and their equivalents. This specification contains multiple inventive concepts; terms such as "preferredly," "according to a preferred embodiment," or "optionally" indicate that the corresponding paragraph discloses an independent concept. The applicant reserves the right to file divisional applications based on each inventive concept. Throughout the text, features introduced by "preferredly" are merely optional and should not be construed as mandatory. Therefore, the applicant reserves the right to abandon or delete relevant preferred features at any time.

Claims

1. An emotion recognition system, comprising: a terminal module integrated with a shooting unit or connected with an independent shooting unit, operated by an individual in a specified place, for obtaining feature data information in the form of pictures and / or videos and realizing uploading and / or publishing of the feature data information, a collection module configured in the specified place, for obtaining feature data information in the form of pictures, voice and / or videos containing one or more individual persons in different positions and angles in the specified place, a processing module in communication connection with the terminal module and the collection module, characterized in that the processing module obtains feature data information containing pictures in a direct and / or indirect manner from the terminal module and the collection module, and determines a corresponding emotion score by recognizing the emotion category of the facial expression in one or more pictures of the feature data information, and the feature data information containing pictures is labeled with geographical attributes and time attributes, the processing module classifies the collected pictures and / or corresponding regions in the pictures based on a first classification rule and / or a second classification rule to assign a corresponding processing weight when analyzing and processing the feature data information containing pictures, the processing module classifies according to the picture source when executing the first classification rule, and the feature data information containing pictures uploaded and / or published by the terminal module is divided into active shooting images, and the feature data information containing pictures obtained by the collection module is divided into passive shooting images, the processing module classifies according to the region where the person in the picture is located when executing the second classification rule, so as to divide any picture into a main image region and a secondary image region, the processing module assigns a lower second processing weight to the main image region than to the secondary image region in the active shooting image, and assigns a higher second processing weight to the main image region than to the secondary image region in the passive shooting image.

2. The affect recognition system of claim 1, wherein, The processing module (300) can perform emotion recognition of each facial expression of any feature data information containing pictures collected based on a trained emotion judgment model to obtain a single image emotion value.

3. The affect recognition system of claim 1, wherein, The processing module (300) can obtain an emotion statistical estimate of all or part of the specified place within a certain period of time based on the label of geographical attributes and time attributes attached to the feature data information.

4. The affect recognition system of claim 1, wherein, The terminal module (100) is integrated or connected with a shooting unit to obtain feature data information containing pictures labeled with geographical attributes and time attributes through the shooting unit, wherein the terminal module (100) can upload the feature data information to the processing module (300) and / or publish it to a social network that can be searched by the processing module (300).

5. The affect recognition system of claim 4, wherein, The processing module (300) can adjust the local information collection frequency according to the distribution of individual persons in the specified place based on the disease condition, wherein the adjustment of the local information collection frequency can at least be achieved by adjusting the collection frequency of the collection unit of the collected feature data information with a specified geographical attribute tag, and / or by adjusting the collection frequency of the processing module (300) in the social network for the feature data information with a specified geographical attribute tag.

6. A method of using the emotion recognition system of any one of claims 1 to 5, characterized in that, Comprise: Obtain feature data information containing picture properties for one or more individual persons in a specified place in at least two ways; Determine the corresponding emotional score by identifying the emotional category of the expression of the person in one or more pictures of the feature data information, Wherein, in the calculation of the emotional score, the picture and / or the corresponding region in the picture can be given a corresponding processing weight based on different classification rules to obtain the emotional statistical estimate of all or part of the specified place in any time period through weight processing.

7. The method of claim 6, wherein, The feature data information containing picture properties is obtained in at least two ways, including active uploading and / or publishing by individual persons and through the collection module (200) arranged in the specified place, wherein the local information collection frequency of any acquisition method can be adjusted according to the distribution of individual persons in the specified place based on the disease condition.

Images