Multi-person real-time processing method combining person identification and behavior identification

Abstract

The invention provides a multi-person real-time processing method combining person identification and behavior identification, and belongs to the field of image processing. The method comprises: S1, obtaining classroom monitoring images in a classroom; S2, extracting a feature map by utilizing the basic neural network model; S3, performing target detection and extraction on the basis of feature map extraction, and performing personnel identity recognition and behavior recognition; S4, performing feature map extraction and rearrangement on each target by using ROI-Pooling operation, and selecting the target with the maximum probability; and S5, inputting the target with the maximum probability into the time sequence information processing LSTM network, and performing personnel identity recognition and behavior recognition optimization in combination with the time sequence information. According to the invention, real-time joint detection and identification are carried out on identity information and behavior information of all persons in a classroom scene with a large number of shielded and mobile persons.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a multi-person real-time processing method combining person recognition and behavior recognition. Background technique [0002] At present, the monitoring and statistics of student classroom behavior refers to allowing the machine to detect and understand the identity information and behavior information of each student in the classroom, and further conduct statistical analysis of each student's behavior to provide quantitative analysis and guidance for their learning situation. Group behavior detection, analysis and monitoring has great application potential in the field of K12 education. Monitoring the quantitative analysis of each student's learning in the classroom and timely feedback to teachers and parents are currently hot research issues in K12 education. [0003] Simultaneously identifying the identity information and behavior information of each student in the classroom ...

AI Technical Summary

Problems solved by technology

[0003] Simultaneously identifying the identity information and behavior information of each student in the classroom based on the surveillance video is a relatively difficult problem in the current image processing field, especially in the classroom scene: there are many people, serious occlusion, and complex environment. Currently there is no processing method capable of simultaneous person identification and behavior identification

Images