Which is the Better Teacher Action? A New Ranking Model and Dataset
Ming Fang, Xinning Du, Qi Liu, Yunpeng Zhou, Qiwen Liang, Shuhua Liu
Abstract
Teachers as leaders of classroom teaching, can enhance students’ learning interest by effectively using body language. Consequently, the quality of teachers’ actions is one of the critical factors influencing the teaching effect. Teachers can find their shortcomings and improve their teaching skills by watching high-quality teaching actions. However, the lack of datasets for teacher action quality assessment has severely hindered its development. To address this issue, this paper constructs the first Teacher Action Quality Ranking dataset (TAQR) and introduces the ranking problem into the field of education for the first time. This dataset comprises 1,200 video samples, covering 9 subjects and 4 kinds of typical teacher actions, demonstrating rich subject diversity. It is currently the largest dataset for ranking teacher action quality. Based on this dataset, a Teacher Action Quality Ranking model (TAQRM) is proposed to select high-quality teacher actions, thereby assisting teachers to improve their teaching behaviors. Experimental results demonstrate the effectiveness of the model in this task. The dataset and code can be openly obtained at https://github.com/MingZier/TAQR-Dataset.
BibTeX
@inproceedings{icassp2024_whichisthebetter,
title = {Which is the Better Teacher Action? A New Ranking Model and Dataset},
author = {Ming Fang and Xinning Du and Qi Liu and Yunpeng Zhou and Qiwen Liang and Shuhua Liu},
booktitle = {ICASSP 2024},
year = {2024}
}