基于3D卷积神经网络的考生异常行为识别研究

Study on recognition of candidate's abnormal behavior based on 3D convolutional neural network

本论文将卷积神经网络应用到的考生异常行为检测的智能视频监控系统中去。当输入多维图像时,算法更为显著,只用直接进行图像输入,还能够使考生异常行为识别的准确性得到提升。

In this thesis, the convolutional neural network is applied to the intelligent video surveillance system for the abnormal behavior detection of candidates. When multi-dimensional images are input, the algorithm is more significant, and only the image input is directly performed, and the accuracy of the candidate's abnormal behavior recognition can be improved.