A Teacher Action Quality Assessment Method Based on Label Constraint Strategy

Teacher’s actions significantly impact teaching effectiveness. Assessing these actions can help teachers to identify shortcomings and improve skills. However, the lack of datasets for assessing teacher action quality has hindered progress in this field. Therefore, this paper first constructs the Teacher Teaching Action Quality Assessment (TTAQA) Dataset, which includes 4 common teacher actions in real classroom, with a total of 2254 samples. In addition, this study proposes a teacher action scoring model based on Label Constraint Strategy (LCS). This model aims to construct a constraint loss by maximizing the similarity between the same class samples and the difference between the different class samples, thereby improving the performance of the model. This is also the first time that the concept of label contrastive learning has been introduced into action quality assessment tasks. A large number of experimental results demonstrate the effectiveness of the LCS method and the rationality of TTAQA dataset.