70 results about "Microexpression" patented technology

The invention provides a method of micro facial expression detection based on a facial action coding system (FACS). The method has a main content of a visual CNN filter, network architecture and training, migration learning and micro facial expression detection. The method comprises steps: firstly, a robust emotion classification framework is built; the provided network learning model is analyzed; the provided network training filter is visualized in different emotion classification tasks; and the model is applied to micro facial expression detection. The recognition rate of the existing method in micro facial expression detection is improved, the strong correlation between features generated by an unsupervised learning process and action units for a facial expression analysis method is presented, the FACS-based function generalization ability in aspects of providing high-precision score cross data and cross mission is verified, the micro facial expression detection recognition rate is improved, the facial expression can be recognized more accurately, the emotion state is deduced, the effectiveness and the accuracy of application in various fields are improved, and development of artificial intelligence is pushed.

The invention discloses a face microexpression recognition method and relates to a method for recognizing graphs with electronic equipment. The method comprises the steps that after a face microexpression image sequence is preprocessed, the Birnbaum-Saunders distribution curve is used for building a regression model, then the CBP-TOP algorithm is used for extracting dynamic space-time texture features of the face microexpression sequence, and finally a classier is used for training and prediction. The defect that an existing face microexpression recognition method is sensitive to small features such as bright points, edges and white noise of a face microexpression image, thereby being low in recognition performance is overcome.

The invention provides a microexpression-based credit authorization method. The microexpression-based credit authorization method includes the steps: acquiring a credit microexpression sample set, andconstructing a microexpression fraud identification model according to the credit microexpression sample set; when receiving a credit authorization instruction, acquiring an original video stream ofapplicant credit Q&A (Questions and Answers), wherein the original video stream includes microexpressions of the applicant credit Q&A process; inputting the original video stream into the microexpression fraud identification model to perform microexpression identification, and obtaining a microexpression identification result; and according to the microexpression identification result, generatingcorresponding credit decision-making suggestion information. The invention also provides a microexpression-based credit authorization device, equipment and a readable storage medium. The microexpression-based credit authorization method uses the microexpression fraud identification model to analyze the microexpressions of a credit applicant to determine the real feeling of the applicant to determine whether the applicant tell lies so as to detect fraud, thus reducing the workload of artificial authorization, and being conductive to improving efficiency and accuracy of credit authorization.

Disclosed is a driving dangerous mood alert method including that follow steps, S1: acquiring a training sample set, S2, preprocess that micro-expression image, S3, using the improved convolution neural network to train the preprocessed microexpression images in the microexpression training sample set, S4: Microexpression image recognition: A driver's face video sequence is captured in real time,S5, judging whether the micro-expression type of the micro-expression image in S4 is the micro-expression corresponding to the driving dangerous emotion or not, if so, proceeding to S6, wherein the micro-expression type is the micro-expression corresponding to the driving dangerous emotion; if not, proceeding to step S7; S6, voice prompt that driver of the mood correspond to the current micro expression, prompting the driver to adjust the mood and steadily drive; S7: No reminders.

The invention provides a multi-view micro-expression recognition method and device, a storage medium and a computer device. The method comprises the following steps: obtaining facial expression data of a target expression of a user, wherein the facial expression data includes figure expression images of a target expression at multiple angles; inputting the expression data of the user into a plurality of preset microexpression recognition models to obtain a set of expression classification probabilities corresponding to each microexpression recognition model, wherein the set of expression classification probabilities comprises a matching probability between the expression data and different expression classifications; calculating an average value of matching probabilities of the same expression classification in each expression classification probability set; determining the expression classification corresponding to the target expression according to the average value of the matching probability of the expression classification. This method can quickly and accurately realize multi-angle multi-pose micro-expression recognition, and meet the needs of multi-angle multi-pose micro-expression recognition.

The invention relates to a micro-expression recognition method based on active migration learning. The method comprises the following steps: (1) extracting micro-expression and micro-expression features; (2) establishing and solving a micro-expression active migration learning problem; (3) recognizing a microexpression. Based on the internal relation between macro expressions and micro expressions, a bridge between macro expressions and micro expressions is constructedthrough an asymmetric linear translator. Less labeled samples are adopted during the initial stage of the active learning. By means of the above translator, the supervision information of the macro-expression domain can be utilized by micro expressions in thetransformation domain. As a result, a high-quality sample can be selected for the active learning in the micro-expression domain. After the sample is manually marked, the sample is added to an existing training set. Therefore, a more effective classifier is obtained through training.

The invention relates to a micro-expression recognition method based on macro-expression knowledge transfer, comprising the following steps: (1) partitioning the expression and the micro-expression; (2) performing feature extraction of facial expression and microexpression, performing LBP feature extraction and optical flow feature extraction; (3) constructing a micro-expression recognition modelof macro-expression knowledge transfer, that is, mapping a specific class of expression and micro-expression learning, and projecting the expression and micro-expression onto multiple common discriminant subspaces; (4) classifying and recognizing the microexpression by a nearest neighbor classifier based on Euclidean distance. On the one hand, multi-feature learning can combine the characteristicsof different features to achieve the best recognition results. On the other hand, multi-task learning is to take the feature points as the center to reduce the influence of other irrelevant regions of the face on the experimental results. Feature points here mainly refer to the key points related to facial expression and micro-expression recognition.

The invention discloses a real-time video emotion analysis method and system based on deep learning. The analysis method comprises the following steps of S1, obtaining a training data set; S2, recognizing microexpressions of the training data set through an algorithm based on a deep neural network, performing screening, and outputting 8 kinds of expression predicating values, wherein 8 kinds of expressions comprise a gentle expression, a happy expression, an amazed expression, a sad expression, an angry expression, a disgusted expression, a fear expression and a despised expression; S3, predicating shot human expressions through a heart rate algorithm, and obtaining corresponding heart rate values; and S4, comparing the heart rate values obtained in the step S3 with the expression predicating values obtained in the step S2, and outputting the expressions the same as the heart rate values obtained in the step S3. According to the real-time video emotion analysis method and system basedon deep learning disclosed by the invention, human face recognition in machine vision and an image classification algorithm are applied to detection of microexpressions and the heart rate, recognitionof the microexpressions is realized through the deep learning algorithm, and the real-time video emotion analysis method and system based on deep learning can be applied to the clinical field, the juridical field and the security field.

The invention provides a micro-expression recognition method based on the main direction of optical flow, which comprises the following steps: interpolating a micro-expression sequence formed by framing a micro-expression video; positioning the eyes and nose of each frame, and calculating the position of the middle point of the eyes as the origin of the coordinate axis; calculating the distance dem and dmn from the origin of the coordinate axis of each frame to the center of each eye and the nose; constructing normalized mesh according to dem and dmn; calculating the optical flow direction ofeach pixel region in all adjacent frame grids and clustering to obtain the optical flow main direction; a rectangular characteristic region being formed in the main direction of the optical flow, andthe pixel region in the main direction of the optical flow being assigned with weight coefficients to obtain the optical flow field optimized by adjacent frame grids, and the curves formed by connecting the related data drawing points being smoothed to obtain the global optical flow field; the pixel region in the main direction of the optical flow being assigned with weight coefficients. A training classifier inputs each optical flow characteristic region and compares the sum of similarities between the traversed-level microexpression sequence and the trained microexpression sequence with S toobtain a classification result.

The present application relates to the field of artificial intelligence, in particular to a patient monitoring method, an apparatus, a computer device and a storage medium. The method comprises: receiving a first patient image and identifying the first patient image to obtain a first patient microexpression; Judging whether the first patient microexpression is expressed as a negative emotion; Identifying the first patient image to obtain a patient facial action when the first patient microexpression is expressed as a negative emotion; Acquiring a monitoring event corresponding to the patient'sfacial action and executing the monitoring event. The method can improve the monitoring quality.

The invention, which belongs to the technical field of computer vision, particularly relates to a minimum-error-based feature extraction method for face microexpression sequence. A microexpression sequence is segmented into small space-time blocks and a two-dimensional principal direction vector is searched in each space-time block according to a minimum error rule; and principal directions of all blocks are spliced to obtain vectors twice as large as the number dimension of the blocks, thereby representing the overall microexpression sequence. According to the method, the requirement of the uniform frame number in the traditional algorithm is avoided and thus introduction of the interpolation algorithm is not required. Meanwhile, the extraction speed is fast and the possibility of high-precision microexpression detection is provided.

The invention provides a micro-expression recognition method, a device and a storage medium. The method comprises the following steps: detecting the facial features of a target face image collected inadvance, and obtaining at least five facial feature points of the target face image; according to the facial feature points of the target face image and the preset image partitioning rules, cutting the target face image into blocks to obtain a plurality of blocks, according to the plurality of image blocks, obtaining the microexpression classification result through a convolution neural network model established in advance. According to the obtained facial feature points and preset image partitioning rules, the target face image is processed by slicing, and a plurality of image blocks obtained after slicing are recognized and classified by adopting a convolution neural network, so that the speed and the precision of the micro expression recognition can be effectively improved, and the working efficiency of the micro expression recognition can be greatly improved.

The invention discloses a listening evaluation method based on listener micro-expression, DEVICE, COMPUTER DEVICE AND STORAGE MEDIUM, The listening evaluation method based on the listener's micro-expression comprises the following steps: because for every listener in the lecture, the microexpressions extracted from the collected facial images of the listener belong to the true state of the listener 's heart, it is determined that the preset emotional state corresponding to the micro-expression also belongs to the true emotional state of the listener 's heart, Next, each first evaluation scorecorresponding to the emotional state of the listener is determined, the first overall rating score of the listener is then calculated, Finally, the activity evaluation score of listening activities isdetermined according to the first total evaluation score of all the listeners, Therefore, this kind of scoring method is more reflective of the audience's true evaluation of listening activities, will not be affected by the audience's subjectivity and other reasons, a better understanding of the effect of listening activities, undoubtedly improve the accuracy of the statistics of the audience's evaluation of listening activities.

The invention discloses an emotion analysis method and system. The method comprises the steps of obtaining a physiological signal corresponding to a target user; wherein the physiological signals comprise electroencephalogram signals and electromyogram signals; acquiring facial image information corresponding to the target user; respectively inputting the physiological signal and the facial imageinformation into at least one pre-trained target classification model to obtain a physiological signal recognition result and a micro-expression recognition result corresponding to the target user, and determining an emotion analysis result corresponding to the target user based on the physiological signal recognition result and the micro-expression recognition result. According to the technical scheme provided by the embodiment of the invention, the technical problems of certain errors and relatively high labor cost caused by manually determining the current state information of the target user in the prior art are solved, and the technical effects of quickly and accurately determining the current state of the user and reducing the labor cost are achieved.

The invention provides a method for recognizing facial micro-expression based on video amplification and depth learning. The method comprises the following steps of: using a video amplification technique based on interference cancellation to amplify the motion amplitude of the micro-expression video data; the enlarged video data being divided into video frame images, and all image sequences belonging to micro-expression are extracted according to the micro-expression tags in the data set to form a new data set; facial clipping preprocessing being carried out on the processed video, and all video image sequences being uniformly clipped into 110*110 size gray-scale images; the new data after preprocessing being put into the convolution neural network model and trained to extract the micro-expression feature data to achieve the task of micro-expression recognition. The technical proposal provided by the invention enlarges the amplitude of the expression action through the video amplification operation of eliminating interference to the complete data set, and simultaneously introduces a neural network model for training, thereby effectively improving the accuracy rate of the micro expression recognition on the basis of the full classification of the emotion label.

The invention discloses a micro expression recognition method based on sparse projection learning. The method comprises the steps of: Step 1, collecting a micro expression sample, extracting LBP features P, Q, R of three orthogonal planes of the micro expression, and defining C, D, E as feature optimization variables of three orthogonal planes of XY, XT, YT respectively; constructing an optimization model; 2, setting the initial value and maximum value of iterative counting variable t and n and initializing the regularization parameter kappa, kappa max, and the scale parameter ruho; step 3, initializing the expression (shown in the description), calculating C, and updating T1 and Kappa; If the expression (shown in the description) converges or n> nmax, proceeding to step 4; step 4, initializing the expression (shown in the description), calculating D, and updating T2 and Kappa; and if the expression (shown in the description) converges or n>nmax, proceeding to step 5; step 5, initializing the expression (shown in the description), calculating E, and updating T3 and Kappa; and if the expression (shown in the description) converges or n> nmax, proceeding to step 6; step 6, makingt=t+1, if t <= tmax, returning to step 3, otherwise, outputting C, D, E; step 7, optimizing the LBP features of the three orthogonal plane by optimizing the variables C, D and E to obtain a new fusionfeature Ftest, and predicting the emotion category of the test sample through the trained SVM classifier for the fusion feature Ftest.

The invention discloses a suicide emotion perception method based on multi-mode fusion of voice and micro-expression. The method comprises the following steps: collecting videos and audios by using Kinect with an infrared camera; analyzing image frames and audios in the video by using different methods and converting the image frames and audios into corresponding feature texts; fusing the featuretexts, namely performing dimension reduction processing to obtain fused features; and classifying the fusion features by using a SoftMax activation function, and judging whether the emotion belongs tosuicide emotions or not. According to the invention, the multi-modal data is aligned with the text layer. The text intermediate representation and the proposed fusion method form a frame fusing speech and facial expressions. According to the invention, the dimensionality of voice and facial expressions is reduced, and two pieces of information are unified into one component. The Kinect is used for data acquisition, and the invention has the advantages of being non-invasive, high in performance and convenient to operate.

The invention provides a multi-modal emotion recognition method based on micro-expressions, limb actions and voice, which comprises the following steps: step 1, inputting a facial video of a subject receiving stimulation of a certain signal, and recognizing the micro-expressions; 2, inputting a body video of a subject receiving stimulation of a certain signal, and identifying limb actions; and 3, inputting an audio signal stimulated by a certain signal received by the subject, and recognizing the voice emotion. The micro-expression recognition result in the first step, the limb action recognition result in the second step and the voice emotion recognition result in the third step are fused, and the continuous emotion state of the current subject is judged. According to the method, the emotion recognized by the micro-expression is combined with the emotion recognized by the limb action and the speech emotion, so that the emotional state of the subject can be predicted more accurately. Compared with the prior art, the method has the beneficial effects that the real emotion of a person can be identified more accurately.

The invention discloses a multi-modal emotion recognition method and system fusing voice and micro-expressions, and relates to the technical field of situation recognition. The method comprises the steps of: establishing a voice emotion database and an emotion association function; acquiring voice information and face image information of the same target object at the same time, and extracting emotion representation vocabularies and micro-expression data; obtaining an emotion correlation function and an emotion fluctuation value corresponding to the micro-expression according to a matching result; establishing an emotion recognition network, and decomposing step by step to obtain a plurality of emotion recognition lines; obtaining a corresponding emotion fluctuation value, and establishingan emotion recognition curve; and selecting a qualified emotion recognition line according to a preset fluctuation degree after the emotion fluctuation degree is calculated. According to the invention, the authenticity of the real-time emotion of the target object represented by the voice information and the face image information is enhanced, the probability that the same situation reflects different situations is reduced, the accuracy of the emotion recognition result is improved, and the error of the emotion recognition result is reduced.

The invention discloses a classroom behavior detection method based on an improved Openpose model and facial micro-expressions, and the method comprises the steps: arranging a camera in front of a desk, and detecting the classroom behaviors of students in real time; and recognizing facial information and upper body skeleton information through an artificial intelligence model, taking whether key points can be recognized or not and the distance between the key points as main judgment conditions, serving changes of micro-expressions as auxiliary judgment conditions, if a certain student does notmeet the condition for a period of time, judging that the examination behavior of the student is abnormal. Besides, through the video stream of one class, the possible stage of abnormal behaviors ofstudents is found out and analyzed, and innovation and reform of teaching are realized. Interference factors are reduced through machine vision recognition, equipment is simplified, and meanwhile theinvention further provides a corresponding data analyzing and processing system. According to the invention, the network model is further optimized through adoption of methods such as residual network, weight trimming and the like. According to the invention, self-service classroom behavior detection and feedback are realized, the test efficiency is high, and the accuracy can reach 95%.

The invention discloses a depression tendency recognition method based on multi-modal characteristics of limbs and micro-expressions. The method comprises the following steps: detecting human motion by means of a non-contact measurement sensor Kinect, and generating motion text description; capturing a face image frame by adopting a non-contact measurement sensor Kinect, performing Gabor wavelet and linear discriminant analysis on a face region of interest, performing feature extraction and dimensionality reduction, and then realizing face expression classification by adopting a three-layer neural network to generate expression text description; performing fusion through text description extracted by a fusion neural network with a self-organizing mapping layer and generating information with emotion features; and S4, using a Softmax classifier to classify the feature information generated in the S3 in emotion categories, wherein a classification result is used for evaluating whether the patient has a depression tendency or not. Static body movement and dynamic body movement are considered, and higher efficiency is achieved. Body movement is helpful for identifying the emotion of adepression patient.

The embodiment of the invention provides a virtual reality sleep promoting method and device. The method comprises the steps of: obtaining a preset hypnotic voice, outputting the hypnotic voice through virtual reality, and obtaining a micro-expression of a user after the hypnotic voice is output through a camera device; analyzing the micro-expression of the user, and judging whether the micro-expression of the user is a positive emotion micro-expression or a negative emotion micro-expression according to an analysis result; when the micro-expression of the user is a negative emotion micro-expression, obtaining a similar voice of the hypnosis voice, outputting the similar voice through virtual reality, and obtaining a feedback micro-expression of the user through the camera device; and comparing the negative emotion micro-expression with the feedback micro-expression to obtain a micro-expression difference of the user, and correspondingly adjusting the content of the hypnosis voice according to the micro-expression difference. By adopting the method, targeted adjustment of the hypnotic voice can be performed for different users; and when the user falls asleep, the hypnotic voice is adjusted through the micro-expression change of the user, so that the falling asleep experience of the user is not affected.