A Multi-Modal Dataset for Teacher Behavior Analysis in Offline Classrooms

We introduce a multi-modal dataset (MM-TBA) for teacher behavior analysis in offline classrooms. MM-TBA is a comprehensive multi-modal dataset designed for analyzing teacher behavior in offline classroom settings. MM-TBA consists of three sub-datasets: the Teacher Teaching Action Detection Sub-Dataset, the Teacher Lecture Evaluation Sub-Dataset, and the Teaching Design Sub-Dataset. The Teacher Teaching Action Detection Sub-Dataset includes visual feature files of teachers' teaching actions and action classification annotation files. The Teacher Lecture Evaluation Sub-Dataset comprises the text of teachers' lectures and corresponding evaluation reports. The Teaching Design Sub-Dataset contains the original teaching designs and their respective annotation files. The aim of introducing MM-TBA is to promote scientific research in teacher behavior and cognitive science, provide new research tools for educational science, and foster the development of intelligent education technology. In the teacher action detection sub-dataset, out of the consideration for the protection of the subjects, we have provided the visual feature files for teacher action detection.