| 摘 要: 随着智慧教室的不断普及，采用传统人工评教方法已经不能满足评估教师在这些教室教学视频中教学行为的需要。针对这些教学视频，建立了智能化评价教师教学行为的框架。这个框架首先基于HRNet深度学习网络获得教师人体姿态信息，然后根据教师的姿态信息建立了评价其教学行为的指标，最后采用模糊综合评价的方法实现了对教师教学行为的综合评分。实验表明，本框架的评价结果总体和传统的人工评价结果是一致的。 |
| 关键词: 教师教学行为；深度学习；模糊综合评价；HRNet模型；姿势识别 |
|
中图分类号: TP391.41
文献标识码: A
|
| 基金项目: 本文由广东省本科高校高等教育教学改革项目“电子信息工程专业新工科人才培养的研究与实践”(421)和广东第二师范学院高等教育教学改革项目(2020jxgg01)资助. |
|
| An Evaluation Method for Teaching Behaviors based on Posture Recognition Algorithm |
|
ZHENG Yuhuang
|
(Academic Affairs Office, Guangdong University of Education, Guangzhou 510303, China )
zhyhaa@126.com
|
| Abstract: With the continuous popularity of smart classrooms, traditional manual evaluation methods are far from meeting the needs of evaluating teaching behaviors in teaching videos of smart classrooms. In view of this problem, this paper proposes a framework for intelligently evaluating teaching behaviors. In the framework, teachers' body posture information is first obtained though HRNet (High Resolution Net) deep learning network. Then an index to evaluate their teaching behavior is established according to the teachers' posture information. Finally, comprehensive scoring of the teachers' teaching behavior is achieved by using the fuzzy comprehensive evaluation method. Experiments show that the overall evaluation results of the proposed framework are consistent with traditional manual evaluation results. |
| Keywords: teaching behaviors; deep learning; fuzzy comprehensive evaluation; HRNet model; posture recognition |
