PoseConv3D支持下的课堂行为识别研究

Recognition of Classroom Behavior Based on PoseConv3D Algorithm

随着人工智能相关技术的飞速发展,基于计算机视觉技术的神经网络模型能够自动提取学生在课堂中的行为特征,从而预测出学生的行为类别,为教师改善课堂质量提供数据支持。为了实现对课堂行为的精确识别,本文利用PoseConv3D算法构建了一个课堂行为识别模型。模型采集和剪辑学生的课堂行为视频片段,并对剪辑后的视频样本进行类别标注,构成课堂行为数据集,用PoseConv3D算法对自建数据集进行训练和评估。实验结果表明,模型对学生的课堂行为识别准确率达到了94.8%。

With the rapid development of artificial intelligence related technologies,neural network models based on computer vision technology can automatically extract students'behavioral characteristics in the classroom,thereby predicting their behavioral categories and providing data support for teachers to improve classroom quality.In order to achieve accurate recognition of classroom behavior,this article constructs a classroom behavior recognition model using the PoseConv3D algorithm.The model collects and edits video clips of students'classroom behavior,and classifies the edited video samples to form a classroom behavior dataset.The self built dataset is trained and evaluated using the PoseConv3D algorithm.The experimental results show that the model achieves an accuracy of 94.8%in identifying students'classroom behavior.