Evaluating Quality of College English Classroom Teaching based on Action Recognition using Genetic Algorithm based Multilayer Perceptron

In these days, enrollment of students in education are rapidly increasing so, teaching quality evaluation need to be addressed. English classrooms should have set of quality teaching activities, feedback procedure is a time taking process and for large-scale classroom analysis, it may not achieve better results. In this paper, Genetic Algorithm based Multilayer Perceptron (GA-MLP) is suggested to evaluate quality of classroom English teaching. The proposed GA-MLP is employed using face expression recognition dataset. Then fast furior transform (FFT) based frequency domain features are extracted from dataset images. Followed by an optimal data clustering method to construct a quality index evaluation system with Delphi method. Principle component analysis is used to remove low-key values and retained high key values for feature dimensionality reduction. The MLP constructs a neural network system and index evaluation system is given as input to generative algorithm to optimize parameters of MLP neural networks. The output layer of MLP with rectified activation function classifies teaching as quality teaching and inferiority teaching.