An FFM-based Model for Students' In-class Learning Postures Estimation

In the recent years, education quality monitoring has been attached great importance by Chinese government, and with the promulgation of the development plan for the new generation of artificial intelligence, the application of this technology in education quality monitoring fields has attracted wide attention of scholars. How to apply relevant algorithms of machine vision to classroom quality assessment is the research focus of smart education. In this paper, considering the timeliness of feedback, the stability of algorithms, the accuracy rate and speed in classroom quality assessment, we propose a model called Face Features Matching (FFM), based on face detection networks and feature points matching methods, which innovatively combines machine learning with parameter estimation and numerical methods. Our model is divided into two sub-modules: face feature extraction and optimal parameter estimation. It can be applied successfully to classroom scenarios and become an important part of automated classroom quality assessment.