Data monitoring device for emotion calculation

Abstract

The invention discloses a data monitoring device for emotion calculation. The monitoring device comprises a face recognition emotion analysis system, an APP feedback system and a holographic projection assembly. The face recognition sentiment analysis system comprises a pre-training model of OpenCV, the pre-training model of OpenCV comprises a Haar cascade detector, and the Haar cascade detector is connected with a face monitoring assembly; the APP feedback system selects a back propagation visualization parameter added to real-time guidance, and gives a CNN (Convolutional Neural Network) with only ReLU (Region Local Unit) as an activation function of a middle layer; according to the holographic projection assembly, image information is coded through laser beams, coherence of light waves is controlled during recording and reproduction, and the effect that an observer can see a real three-dimensional image with the depth sense is achieved. Compared with the prior art, the method has the advantages that an emotion analysis model and an artificial intelligence model can be integrated, and interaction is achieved through application programs.

Description

Technical Field

[0001] The present invention relates to the technical field of emotion data monitoring, and specifically to a data monitoring device for emotion computing. Background Art

[0002] The growth of the emotion monitoring market benefits from the continuous innovation and breakthroughs in camera sensors, face detection and recognition technologies, and emotion computing technologies. With the increasing maturity of these technologies, the performance of emotion monitoring products has been significantly improved, and their recognition accuracy, real-time performance, and user adaptability have been greatly enhanced. This provides a broader space for the application of emotion monitoring products in the fields of mental health, education, enterprise employee emotion management, social public safety, etc. Through the integrated high-sensitivity camera sensor, high-precision capture of the user's facial expressions is achieved, and combined with advanced face detection and recognition technologies, the user's emotional changes are accurately identified. The application of emotion computing technology further analyzes and understands these emotional changes, providing personalized emotional feedback and solutions for users. These technological breakthroughs and innovations provide strong impetus for the growth of the emotion monitoring market and open up new possibilities for research and application in related fields. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the above technical defects.

[0004] To solve the above problems, the technical solution of the present invention is: a data monitoring device for emotion computing, the monitoring device includes a face recognition emotion analysis system, an APP feedback system, and a holographic projection component;

[0005] The face recognition emotion analysis system includes a pre-trained model of OpenCV, the pre-trained model of OpenCV includes a Haar cascade detector, and a face monitoring component is connected to the Haar cascade detector, and the face monitoring component reads and identifies the features of the face;

[0006] Based on the above operations, Model One and Model Two are established;

[0007] Model One uses the InceptionV3 architecture to remove the fully connected layer and uses global average pooling;

[0008] Model Two uses the Xception architecture and uses depthwise separable convolution and residual modules to further;

[0009] Based on the above Model One and Model Two, the ADAM optimizer is used for model training, Global AveragePooling is used to completely eliminate any fully connected layer, the softmax activation function graph is applied, and the used model is used to perform real-time classification to reduce the number of parameters for each reduced feature map;

[0010] The described APP feedback system selects the backpropagation visualization parameters added to real-time guidance, and gives a CNN with only ReLU as the activation function of the middle layer;

[0011] The described holographic projection component encodes image information using laser beams and realizes the effect that observers can see real three-dimensional images with a sense of depth by controlling the coherence of light waves during recording and reproduction.

[0012] Furthermore, the ReLU CNN reconstructed image is given by the following formula:

[0013]

[0014] Among them, the guided backpropagation takes the derivative of each element (x, y) of the input image I with respect to the element (i, j) of the feature map fL in layer L.

[0015] Furthermore, the Haar cascade classifier classifies by comparing the Haar-like feature values of faces and non-faces. When the classifier judges it as a face, the output is 1, and when the classifier judges it as a non-face, the output is 0. In each node of each level of the Haar cascade classifier, the AdaBoost algorithm is used to learn a multi-layer classifier with a high detection rate and a low rejection rate.

[0016] Furthermore, the described APP feedback system includes a home page, a monitoring page, and a shooting page. The home page provides monitoring paths, data recording, an emotion assistant, and a relevant information promotion module. Through the home page, users can select different monitoring paths, search for three months of emotion data and the emotion feedback and suggestions of the emotion assistant. The monitoring page provides real-time monitoring images and the three-month emotion change trend of emotions and detection objects. The shooting page can perform emotion calculation through the mobile phone camera.

[0017] Furthermore, the described holographic projection component adopts Heliodisplay projection technology, including a basic unit and a projection unit. The basic unit generates the water vapor screen necessary for displaying images. The projection source unit projects the image into the air.

[0018] The advantages of the present invention compared with the existing technology are as follows:

[0019] The present invention includes a face recognition emotion analysis system, an APP feedback system, and a holographic projection component. It integrates an emotion analysis model + an artificial intelligence model, realizes interaction through an application program, introduces holographic projection technology + digital human into the product, and combines a face recognition algorithm, a multi-modal emotion analysis system with the APP feedback using "EmoGuardian" to form a personalized emotion database for assisting in analyzing the psychological condition of the user and giving corresponding positive feedback. It can be deployed to the cloud to achieve corresponding functions. In terms of hardware, the expected effect is achieved by using fog screen projection technology, and a three-dimensional display is innovatively added based on heliodisplay to provide depth information, enabling the 3D image to interact with the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the flowchart of a data monitoring device for emotion computing according to the present invention Figure 1 。

[0021] Figure 2 is the flowchart of a data monitoring device for emotion computing according to the present invention Figure 2 。 DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The following further describes the specific implementation method of the present invention with reference to the drawings.

[0023] In order to make the content of the present invention easier to be clearly understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention.

[0024] In specific implementation: as Figure 1 and Figure 2 shown, a data monitoring device for emotion computing, the monitoring device includes a face recognition emotion analysis system, an APP feedback system, and a holographic projection component;

[0025] The face recognition emotion analysis system includes a pre-trained model of OpenCV. The pre-trained model of OpenCV includes a Haar cascade detector. A face monitoring component is connected to the Haar cascade detector. The face monitoring component reads and identifies the features of the face.

[0026] Based on the above operations, Model One and Model Two are established;

[0027] Model One uses the InceptionV3 architecture to remove the fully connected layer and uses global average pooling;

[0028] Model Two uses the Xception architecture and further uses depthwise separable convolution and residual modules;

[0029] Based on the above Model 1 and Model 2, the ADAM optimizer is used for model training, Global Average Pooling is used to completely eliminate any fully connected layers, and the softmax activation function graph is applied. The model used will perform real-time classification, reducing the number of parameters for each reduced feature map;

[0030] The APP feedback system selects to add the visualization parameters of backpropagation for real-time guidance, and gives a CNN with only ReLU as the activation function for the middle layer;

[0031] The holographic projection component encodes image information using laser beams. By controlling the coherence of light waves during recording and reproduction, it achieves the effect that observers can see real three-dimensional images with a sense of depth.

[0032] The ReLU CNN reconstructed image is given by the following formula:

[0033]

[0034] Among them, guided backpropagation takes the derivative of each element (x, y) of the input image I with respect to the element (i, j) of the feature map fL in layer L.

[0035] The Haar cascade classifier classifies by comparing the Haar-like feature values of faces and non-faces. When the classifier determines it as a face, the output is 1, and when the classifier determines it as a non-face, the output is 0. In each level of nodes of the Haar cascade classifier, the AdaBoost algorithm is used to learn a multi-layer classifier with a high detection rate and a low rejection rate.

[0036] The APP feedback system includes a home page, a monitoring page, and a shooting page. The home page provides monitoring paths, data recording, an emotion assistant, and a relevant information promotion module. Through the home page, users can select different monitoring paths, search for three months of emotion data, and the emotion feedback and suggestions of the emotion assistant. The monitoring page provides real-time monitoring images and the three-month emotion change trends of emotions and detection objects. The shooting page can perform emotion calculation through the mobile phone camera.

[0037] The holographic projection component adopts Heliodisplay projection technology, including a basic unit and a projection unit. The basic unit generates the water vapor screen necessary for displaying images. The projection source unit projects the image into mid-air.

[0038] In specific use, the face recognition emotion analysis system can perform real-time analysis in video streams or real-time interactions. The system can instantly analyze the emotional responses of individuals and is applicable to various scenarios such as customer service, education, and mental health monitoring. Through long-term recognition and analysis of the same individual, the system can continuously enrich its dataset, adjust parameters in combination with the supervision mechanism, and improve the accuracy of emotional analysis for different individuals. In addition, by uploading video and picture sources from the app, the system supports offline analysis and gives the results of emotional analysis, which are then fed back through the app. In the above process, audio and text inputs are added to the data input, making the input data more abundant and the emotional recognition of users more accurate. At the same time, due to the introduction of audio and text data inputs, users can obtain a better interaction experience. In terms of model training, a training method combining contrastive learning and the Prompt mechanism is adopted. By comparing data of the same modality and different modalities, not only can the emotional calculation accuracy of the same modality be improved, but also the emotions corresponding to different modality inputs can be well matched.

[0039] Among them, for the Heliodisplay projection system, a three-dimensional fog display is added. Based on the traditional fog screen display technology, a three-dimensional fog display is added, and a fog emission module array is set up. The traditional single fog emitter is increased to four separate fog layer flow column emitters, which can quickly reset the fog screen elements according to the imaging coordinate points. A separate linear motion platform is set up to move the emitter module to different depths to create a non-planar fog screen and display the volume data in the real three-dimensional space.

[0040] The above describes the present invention and its implementation manners. Such a description is not restrictive, and what is shown in the drawings is only one of the implementation manners of the present invention. The actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and, without departing from the gist of the present invention creation, design structurally similar ways and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. A data monitoring device for emotion computing, characterized in that: The monitoring device includes a face recognition emotion analysis system, an APP feedback system, and a holographic projection component; The face recognition emotion analysis system includes a pre-trained model of OpenCV. The pre-trained model of OpenCV includes a Haar cascade detector. A face monitoring component is connected to the Haar cascade detector. The face monitoring component reads and identifies the features of a face; Based on the above operations, Model 1 and Model 2 are established; Model 1 uses the InceptionV3 architecture to remove the fully connected layer and uses global average pooling; Model 2 uses the Xception architecture and utilizes depthwise separable convolution and residual modules to further... Based on the above Model 1 and Model 2, the ADAM optimizer is used for model training. Global Average Pooling is used to completely eliminate any fully connected layer. The softmax activation function graph is applied. The used models are classified in real time to reduce the number of parameters for each reduced feature map; The APP feedback system selects the backpropagation visualization parameters added to the real-time guidance and gives a CNN with only ReLU as the activation function for the intermediate layer; The holographic projection component encodes image information using laser beams. By controlling the coherence of light waves during recording and reproduction, the effect is achieved that the observer can see a real three-dimensional image with a sense of depth.

2. The data monitoring device for emotion computing according to claim 1, characterized in that: The ReLUCNN reconstructed image is given by the following formula: where the guided backpropagation takes the derivative of each element (x, y) of the input image I with respect to the element (i, j) of the feature map fL in layer L.

3. The data monitoring device for emotion computing according to claim 1, wherein: The Haar cascade classifier classifies by comparing the Haar-like feature values of faces and non-faces. When the classifier determines it is a face, the output is 1. When the classifier determines it is not a face, the output is 0. In each node of each level of the Haar cascade classifier, the AdaBoost algorithm is used to learn a multi-layer classifier with a high detection rate and a low rejection rate.

4. A data monitoring device for emotion computing according to claim 1, characterized in that: The APP feedback system includes a home page, a monitoring page, and a shooting page. The home page provides monitoring paths, data recording, an emotion assistant, and a relevant information promotion module. Through the home page, users can select different monitoring paths, search for three months of emotion data and the emotion feedback and suggestions of the emotion assistant. The monitoring page provides real-time monitoring images and the three-month emotion change trend of the emotion and the detection object. The shooting page can perform emotion calculation through the mobile phone camera.

5. The data monitoring device for emotion computing according to claim 1, characterized in that: The holographic projection component adopts Heliodisplay projection technology and includes a basic unit and a projection unit. The basic unit generates the water vapor screen necessary for displaying the image. The projection source unit projects the image into the air.

Images