嵌入式平台基于人脸检测与跟踪的实时人数统计方法

Real-time People Counting Method Based and Face Detection and Tracking in Embedded Platform

提出了一种基于嵌入式平台的实时人数统计方法:使用基于肤色过滤的多层级联快速人脸检测算法进行人脸检测,对检测到的人脸区域使用核相关滤波器(Kernelized Correlation Filters,KCF)跟踪,增加校准模块进一步修正跟踪结果。结合跳帧检测与局部滤除的方式从而在兼顾检测速度以及精度的前提下实现了嵌入式平台下的实时人数统计。经过实验测试,该方法具有较快的检测与统计速度,在嵌入式平台上取得了预期的效果,实验场景下人数统计正确率可达96%以上。

In this paper,a real-time People Counting method based on embedded platform is proposed：using the multilevel cascade fast face detection algorithm based on skin color filtering for face detection and the KCF method to track the detected face area,adding the calibration module to further correct the tracking result.Combined with the method of jump frame detection and local filtering,so as to realize the real-time statistics of the embedded platform under the premise of taking into account the detection speed and precision.