Deep learning is changing the traditional research method in many fields,which not only include image classification,speech recognition,text classification etc.,but also include some more specific industries,such as researches of power equipment detection and public security maintenance,and the empowerment of deep learning and neural network provide better accuracy and efficiency of these kinds of work.Among these research fields,education is no exception.This paper focuses the work on the classification of classroom instructional behavior,within which a summary of current research is conducted firstly,then the mainstreammethods and algorithms that are used in this field are introduced and developed based on the framework of TensorFlow and the programming language of Python.With the training texts extracted from the class videos,an experiment is conducted to compare the accuracy and some other metrics among the different models used to classify the classroom instructional behavior.In the end of this paper,a summary of the work is made to be used as reference by future researches. |