Fast face detection integrating motion energy into a cascade-structured classifier

Authors:
Yafeng Deng;Guangda Su;Jun Zhou;Bo Fu
Affiliations:
Department of Electronic Engineering, Tsinghua University, Beijing, China;Department of Electronic Engineering, Tsinghua University, Beijing, China;Department of Electronic Engineering, Tsinghua University, Beijing, China;Department of Electronic Engineering, Tsinghua University, Beijing, China
Venue:
ICMLC'05 Proceedings of the 4th international conference on Advances in Machine Learning and Cybernetics
Year:
2005

Citing 3
Cited 0

Neural Network-Based Face Detection

IEEE Transactions on Pattern Analysis and Machine Intelligence
Improved boosting algorithms using confidence-rated predictions

COLT' 98 Proceedings of the eleventh annual conference on Computational learning theory
Detecting Faces in Images: A Survey

IEEE Transactions on Pattern Analysis and Machine Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose a fast and robust face detection method. We train a cascade-structured classifier with boosted haar-like features which uses intensity information only. To speed up the process, we integrate motion energy into the cascade-structured classifier. Motion energy can represent moving the extent of the candidate regions, which is used to reject most of the candidate windows and thus accelerates the evaluation procedure. According to the face presence situation, we divide the system state into three modes, and process input images with an intensity detector, or motion integrated dynamic detector, or else keep the pre-results. Since motion energy can be computed efficiently, processing speed is greatly accelerated. Furthermore, without depending on any supposed motion model, the system is very robust in real situations without the limitation of moving patterns including speed and direction.