Student behavior recognition based on multi-view learning via mask RCNN and its application in education management

Abstract

Behavior recognition technology has a wide range of applications, but only refers to the field of security applications, due to the intrusion of anti-terrorism issues, governments, security departments, enterprise organizations and institutions around the world have increased security vigilance. The traditional behavior recognition technologies cannot effectively recognize the face in all directions, because of the face occlusion, incomplete feature extraction, noise and other factors. Therefore, this paper proposes a novel student behavior recognition combined with multi-view learning and mask RCNN (Region-based convolutional neural network. In order to improve the accuracy of occlusion behavior recognition, occlusion segmentation network is used to extract occlusion information, and mask generator generates mask according to occlusion information to hide the damaged features. This paper also proposes a channel-space attention fusion network for feature extraction. The feature norm is incorporated into the loss function as an index of image quality. The experimental results on public data sets show that the proposed method can effectively reduce the interference of unrecognized images and improve the performance of low quality behavior recognition.

Keywords

• behavior recognition
• mask rcnn
• attention feature fusion
• feature extraction