说明: 机器学习专门研究计算机怎样模拟或实现人类的学习行为,以获取新的知识或技 能,为包括人脸识别领域的人工智能的发展做出了极大的贡献。本文探索如何应用机器学习中的一些技术,使计算机更好地完成人脸识别领域中的人脸检测和人脸验证。
在人脸检测方面,针对如何快速、准确地检出人脸的问题,基于DLIB中特征模型,可以快速提取检测出人脸,并且提取出人脸特征点,主要作用就是快速检测定位人脸。在训练过程中,引入了ResNet机器学习算法,该算法采用多层卷积神经网络结构,对人脸进行多层特征提取和描述,得到人脸特征描述符。通过测试可以证实了上述方法能够检测和识别人脸。(In terms of face detection, for the problem of how to detect faces quickly and accurately, based on the feature model in DLIB, you can quickly extract and detect faces, and extract face feature points, the main role is to quickly detect and locate faces. In the training process, the VGG-FACE machine learning algorithm is introduced. This algorithm uses a multi-layer convolutional neural network structure to perform multi-layer feature extraction and description on the face to obtain the face feature descriptor. Tests can confirm that the above method can detect and recognize human faces.)