基于Keras卷积神经网络的多人脸识别抗干扰研究

Research on Anti-jamming of Multi-face Recognition Based on Keras Convolutional Neural Network

人脸识别的训练预测模型是多样而复杂的,影响人脸识别准确率和稳定性的因素也很多。人脸识别的抗干扰设计是构建人脸识别模型不可忽视的重要内容。通过获取更高质量的人脸图像数据源,选择效果更好的人脸识别优化器以及部分超参数的调整,来提高对大量人脸数据进行处理的能力。利用卷积神经网络减少人工干预,提高特征提取的算法精度,从而提高多人脸识别的精度。

There are many complicated face recognition training prediction models,with many factors influencing the accuracy and stability of face recognition.Anti-jamming technology research is essential to construct the face recognition models,which aim to get the quality data,selects face recognition optimizer and adjusts other parameters in order to improve the face data processing.Also,convolutional neural network is applied to reduce human intervention and improve the accuracy of feature extraction,realizing the multi-face recognition.