一种改进的MTCNN和FaceNet人脸检测及识别算法研究

An Improved Face Detection and Recognition Algorithm Based on FaceNet and MTCNN

MTCNN和FaceNet的人脸识别算法是较为常用的算法,但由于各种因素使得识别准确率较低。为了保障基于MTCNN和FaceNet的人脸检测识别系统更好地解决传统安保人工监控模式带来的资源浪费及效率低下等问题,提出了一种改进的MTCNN和FaceNet人脸检测及识别算法。在现有的MTCNN和FaceNet人脸检测及识别算法的基础上,针对夜晚图像出现的大量噪点,使用两次中值滤波进行快速去噪,降低噪点导致的误检率,从而提高图片识别准确率;并将改进前后的算法分别对同个人脸数据集进行检测识别,比较改进前后算法的准确率和召回率,可验证改进后算法的优化有效性。

Face recognition algorithms based on MTCNN and FaceNet are commonly used,but their recognition accuracy is low due to various factors.In order to ensure that the face detection and recognition system based on MTCNN and FaceNet can better solve the problem of resource waste and inefficiency brought by the traditional security manual monitoring mode,this paper proposes an improved face detection and recognition algorithm based on MTCNN and FaceNet.On the basis of the existing FACE detection and recognition algorithms of MTCNN and FaceNet,two median filters are used to quickly de-noise images at night to reduce the false detection rate caused by noise points and improve the image recognition accuracy.Finally,the algorithm before and after the improvement is respectively used to detect and identify the same personal face data set.By comparing the accuracy rate and recall rate of the algorithm before and after the improvement,the optimization effectiveness of the improved algorithm is verified.