A New Method of Classroom Behavior Recognition Based on WS-FC SLOWFAST

A New Method of Classroom Behavior Recognition Based on WS-FC SLOWFAST

Damin Ding (Shanghai University of Engineering Science, China), Yueyang Zhao (Shanghai University of Engineering Science, China), Jingru Zhang (Universiti Sains Malaysia, Malaysia & Shanghai University of Engineering Science, China), Jin Liu (Shanghai University of Engineering Science, China), Jun Liu (Shanghai University of Engineering Science, China), Haima Yang (University of Shanghai for Science and Technology, China), Hongli Shan (Shanghai University of Engineering Science, China), and Zhiwen Zhou (Ministry of Agriculture and Rural Affairs, China)
Copyright: © 2025 |Pages: 28
DOI: 10.4018/IJGCMS.371423
Article PDF Download
Open access articles are freely available for download

Abstract

With the growing integration of deep learning and educational informatization, applying artificial intelligence to classroom behavior analysis has garnered significant attention. This article specifies 14 types of classroom behaviors and their classification criteria. By clipping and frame extraction from surveillance videos, target detection, manual annotation, temporal association, and other operations, a multi-label behavior dataset was created. This article also proposes a Weakly supervised fine-grained classification SlowFast SlowFast behavior recognition algorithm, which improves the accuracy of recognizing small difference classroom behaviors from an intra-class classification perspective. By using attention-guided local feature enhancement in the path, weakly supervised fine-grained classification of behavior target local features was achieved. Experimental results showed the algorithm improves behavior recognition accuracy by 4%-11% for specific behaviors and 5.75% overall, contributing to teaching quality evaluation systems.
Article Preview
Top

Introduction

In the era of the knowledge economy, education and teaching activities on campus have gradually become the focus of social attention and play a vital role in the development of society (Akila et al., 2024). The performance of students’ learning behaviors in the classroom is crucial for teaching and assessment (Liu et al., 2021). Automated classroom behavior measurement methods have become a research hot spot in the field of education informatization because of their highly automated, real-time, and efficient characteristics (Jisi et al., 2021; Lin et al., 2021; Wu et al., 2021).

The State Council of China issued the document “China Education Modernization 2035” in 2019, emphasizing accelerating the transformation process of education informatization and education modernization. It mainly proposed enhancing the development of smart campuses and classrooms, along with integrating intelligent platforms for teaching, management, and services (Xu et al., 2020).

With the rapid development of artificial intelligence technology, deep learning technology has also achieved remarkable success in the fields of image and speech recognition and natural language processing. A primary aim of educational informatization reform is to explore how information technology can contribute to the innovation and transformation of teaching. Combining artificial intelligence technology with campus environments and carrying out special research will help promote the application and innovation of educational modernization.

A study based on activity theory examined the digital teaching activities of anatomy faculty at an Australian university. It revealed several methodological dilemmas experienced by the researcher (Colasante, 2024). This method improves the quality and efficiency of education in complex teaching environments by introducing deep learning technology to automatically analyze classroom videos. It includes the addition of a feature pyramid and convolutional block attention module for comparative experiments to assess teacher and student behaviors and interactions (Xu, T. et al., 2023). A positioning method for facial feature points was developed based on a deep convolutional neural network and cascading, and the head poses and facial expressions were analyzed and recognized (Huang et al., 2020).

In view of the problem that the accuracy of the system’s recognition of students’ classroom behavior is generally low, an improved detection model based on You Only Look Once version 5 (YOLOv5) has been proposed to improve the convergence speed of the prediction box (Zheng et al., 2022). An improved classroom learning behavior recognition algorithm based on You Only Look Once version 8 (YOLOv8n) has been proposed for the small object problem of classroom learning behaviors for back-row-students. A study expanded a tiny object detection layer to detect small targets better (Liu et al., 2024). A kind of class action recognition method on the basis of graph convolution has been proposed, which takes classroom video as input and recognizes student actions through a skeleton extraction module, followed by a feature extraction module based on graph convolution and a zero-shot feature classification module, and this model has achieved significant performance with an accuracy of 60.67% in collected classroom dataset (Shi et al., 2021). Another model significantly enhanced the model’s recognition performance in complex, occluded backgrounds by integrating YOLOv5 with the Channel Attention (CA) attention mechanism, achieving a maximum mean Average Precision (mAP) of 82.1% in complex classroom environments, which surpassed Faster Region-based Convolutional Neural Network (Faster R-CNN) and the original YOLOv5 by 5.2 and 4.6 percentage points respectively (Jia & He, 2024).

Complete Article List

Search this Journal:
Reset
Volume 17: 1 Issue (2025)
Volume 16: 1 Issue (2024)
Volume 15: 1 Issue (2023)
Volume 14: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 13: 4 Issues (2021)
Volume 12: 4 Issues (2020)
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing