Face tracking system based on neural network and optical flow method and tracking method thereof

The invention provides a super-real-time face stable tracking system based on a deep neural network and an optical flow method and a tracking method thereof. The method comprises the following steps: determining the first frame face frame through a deep neural network; comparing a current picture frame with the previous frame through an optical flow method to obtain a current frame face frame; using the deep neural network to verify whether an image obtained by the optical flow method is a human face or not every several frames; and performing tracking and verification, and if any stage fails, repeatedly executing a face detection routine until a face appears, and entering a tracking routine. According to the tracking system, super-real-time long-acting stable face tracking can be realized, so that the actual application effect of a face detection and tracking method in various fields is improved. 本发明提供了一种基于深度神经网络和光流法的超实时人脸稳定追踪系统和追踪方法，通过深度神经网络确定首帧人脸框；通过光流法，对当前画面帧和上一帧做对比，得到当前帧人脸框；每隔几帧使用深度神经网络验证光流法得到的图像是否是人脸；追踪