The Advisable Technology of Key-Point Detection and Expression Recognition for an Intelligent Class System

Abstract

The intelligent classroom system has a very wide range of application scenarios in modern education. With the continuous breakthrough of artificial intelligence technology, a realtime intelligent system that can judge students' performance in class has a large demand market. In this paper, the real-time detection technology needed by the smart classroom system is studied from the aspects of behavior detection and expression recognition. In terms of behavior detection, this paper combines the two key point detection technologies CPM (Convolutional Pose Machines) and CMU (Carnegie Mellon University) OPENPOSE [1] [2]. In terms of expression recognition, in order to detect the position of the face in real time, this paper studies several popular target detection algorithms and improves the traditional CNN network, and proposes a real-time image classification network architecture [8].