Facilitating Intelligent Evaluation and Analysis of Teachers’ Teaching Behaviors Based on Improved Yolov8 Algorithm

Teachers' classroom behavior analysis is a crucial component of teaching analysis and evaluation. With the deep integration of Artificial Intelligence technology and the field of education and teaching, this paper investigates the application of object detection techniques to delve into teacher behavior data, optimize the traditional mode of teaching analysis, and ensure teaching evaluations are more objective and fairer. Integrating the ResT into YOLOv8 algorithm, ResT is an efficient multi-scale visual Transformer structure, which is based on the design structure of ResNet to capture the feature information of the object on different scales in a phased and fine-grained manner, and models the global dependence of the image relationship to enhance the feature extraction ability of the model on the target object, while reducing the model's attention to the background information of the image to achieve high-precision teachers’ behavior detection for Remote Education. Extensive experiments demonstrate that, relative to the benchmark model YOLOv8, the algorithm presented in this paper achieves 93.8%, 92.8%, 95.3%, and 85.3% in Precision, Recall, mAP50, and mAP50-95 metrics respectively on the self-constructed teachers' classroom behavior dataset, which is an improvement of 3.5%, 2.9%, 3.1%, and 2.7%, respectively.