Abstract: Provided are a facial expression recognition method and system combined with an attention mechanism. The method comprises: detecting faces comprised in each video frame in a video sequence, and extracting corresponding facial ROIs, so as to obtain facial pictures in each video frame; aligning the facial pictures in each video frame on the basis of location information of facial feature points of the facial pictures; inputting the aligned facial pictures into a residual neural network, and extracting spatial features of facial expressions corresponding to the facial pictures; inputting the spatial features of the facial expressions into a hybrid attention module to acquire fused features of the facial expressions; inputting the fused features of the facial expressions into a gated recurrent unit, and extracting temporal features of the facial expressions; and inputting the temporal features of the facial expressions into a fully connected layer, and classifying and recognizing the facial expressions.

Abstract: The present disclosure provides a non-contact fatigue detection system and method based on rPPG. The system and method adopt multi-thread synchronous communication for real-time acquisition and processing of rPPG signal, enabling fatigue status detection. In this setup, the first thread handles real-time rPPG data capture, storage and concatenation, while the second thread conducts real-time analysis and fatigue detection of rPPG data. Through a combination of skin detection and LUV color space conversion, rPPG raw signal extraction is achieved, effectively eliminating interference from internal and external environmental facial noise; Subsequently, an adaptive multi-stage filtering process enhances the signal-to-noise ratio, and a multi-dimensional fusion CNN model ensures accurate detection of respiration and heart rate.

Abstract: A pulse ablation catheter for treating pulmonary arterial hypertension includes an ablation catheter part and an energy platform, where the ablation catheter part includes a movable outer sheath tube, an ablation catheter, a functional guide core, a pulse ablation part and an operating part; the operating part controls the pulse ablation part to slide axially on the functional guide core in a center of the ablation catheter, and controls strip-shaped arches in a middle of the pulse ablation part to arch or fold in a circumferential direction so as to perform pulse ablation through pulse electrode sheets and the energy platform. The pulse ablation catheter for treating pulmonary arterial hypertension of the application performs pulse ablation through the pulse ablation catheter with controllable frequency and ablation end shape to treat pulmonary arterial hypertension.

Abstract: Disclosed are a non-contact fatigue detection method and system. The method comprises: sending a millimeter-wave (mmWave) radar signal to a person being detected, receiving an echo signal reflected from the person, and determining a time-frequency domain feature, a non-linear feature and a time-series feature of a vital sign signal; acquiring a facial video image of the person, and performing facial detection and alignment on the basis of the facial video image, for extracting a time domain feature and a spatial domain feature of the person's face; fusing the determined vital sign signal with the time domain feature and the spatial domain feature of the person's face, for obtaining a fused feature; and determining whether the person is in a fatigued state by the fused feature. By fusing the two detection techniques, the method effectively suppressing the interference of subjective and objective factors, and improving the accuracy of fatigue detection.

Abstract: Disclosed are a millimeter-wave (mmWave) radar-based non-contact identity recognition method and system. The method comprises: emitting an mmWave radar signal to a user to be recognized, and receiving an echo signal reflected from the user; performing clutter suppression and echo selection on the echo signal, and extracting a heartbeat signal; segmenting the heartbeat signal beat by beat, and determining its corresponding beat features; and comparing the beat features of the user with the beat feature sets of a standard user group; if the beat features of the user matches one of the beat feature set in the standard user group, the identity recognition being successful; otherwise, being not successful. According to the method, the use of a heartbeat signal for identity recognition has high reliability, and the use of an mmWave radar technology for non-contact identity recognition has high flexibility and accuracy.

Abstract: The present invention discloses construction method and system of a descriptive model of classroom teaching behavior events. The construction method includes steps as the followings: acquiring classroom teaching video data to be trained; dividing the classroom teaching video data to be trained into multiple events according to utterances of a teacher by using a voice activity detection technology; and performing multi-modal recognition on all events by using multiple artificial intelligence technologies to divide the events into sub-events in multiple dimensions, establishing an event descriptive model according to the sub-events, and describing various teaching behavior events of the teacher in a classroom. The present invention divides a classroom video according to voice, which can ensure the completeness of the teacher's non-verbal behavior in each event to the greatest extent.

Abstract: Provided are a facial expression recognition method and system combined with an attention mechanism. The method comprises: detecting faces comprised in each video frame in a video sequence, and extracting corresponding facial ROIs, so as to obtain facial pictures in each video frame; aligning the facial pictures in each video frame on the basis of location information of facial feature points of the facial pictures; inputting the aligned facial pictures into a residual neural network, and extracting spatial features of facial expressions corresponding to the facial pictures; inputting the spatial features of the facial expressions into a hybrid attention module to acquire fused features of the facial expressions; inputting the fused features of the facial expressions into a gated recurrent unit, and extracting temporal features of the facial expressions; and inputting the temporal features of the facial expressions into a fully connected layer, and classifying and recognizing the facial expressions.

Abstract: The invention provides a classroom cognitive load detection system belonging to the field of education informationization, which includes the following. A task completion feature collecting module records an answer response time and a correct answer rate of a student when completing a task. A cognitive load self-assessment collecting module quantifies and analyzes a mental effort and a task subjective difficulty by a rating scale. An expression and attention feature collecting module collects a student classroom performance video to obtain a face region through a face detection and counting a smiley face duration and a watching duration of the student according to a video analysis result. A feature fusion module fuses aforesaid six indexes into a characteristic vector. A cognitive load determining module inputs the characteristic vector to a classifier to identify a classroom cognitive load level of the student.

Abstract: The disclosure discloses a fingerprint positioning method in a smart classroom, which is specifically: firstly, performing Gaussian filtering and taking the average value on a wireless signal strength value RSSI in the fingerprint database; then finding the neighbor point closest to the signal strength of the to-be-measured point; finally, the Euclidean distance is used as the weight reference, and the weighted center of mass is obtained for the nearest neighbor points. The weight index is introduced as an index of the weight, and the coordinates of the to-be-tested node are obtained. The disclosure has a higher positioning accuracy, smaller positioning error fluctuations and greater environmental adaptability.

Abstract: The disclosure discloses a fingerprint positioning method in a smart classroom, which is specifically: firstly, performing Gaussian filtering and taking the average value on a wireless signal strength value RSSI in the fingerprint database; then finding the neighbor point closest to the signal strength of the to-be-measured point; finally, the Euclidean distance is used as the weight reference, and the weighted center of mass is obtained for the nearest neighbor points. The weight index is introduced as an index of the weight, and the coordinates of the to-be-tested node are obtained. The disclosure has a higher positioning accuracy, smaller positioning error fluctuations and greater environmental adaptability.

Abstract: The invention provides a classroom cognitive load detection system belonging to the field of education informationization, which includes the following. A task completion feature collecting module records an answer response time and a correct answer rate of a student when completing a task. A cognitive load self-assessment collecting module quantifies and analyzes a mental effort and a task subjective difficulty by a rating scale. An expression and attention feature collecting module collects a student classroom performance video to obtain a face region through a face detection and counting a smiley face duration and a watching duration of the student according to a video analysis result. A feature fusion module fuses aforesaid six indexes into a characteristic vector. A cognitive load determining module inputs the characteristic vector to a classifier to identify a classroom cognitive load level of the student.