Fast and accurate face detection by sparse Bayesian extreme learning machine

Chi-Man Vong, Keng Iam Tai, Chi-Man Pun, Pak-Kin Wong

Abstract

Real-time face detection is an important research topic in computer vision and pattern recognition. One of the effective methods in face detection is model-based approach which employs neural network technique for the construction of classification model. Relevant techniques such as support vector machines are fast in training an accurate model which is, however, relatively slow in execution time. The reason is due to the large size of the constructed model. In this paper, the main contribution is to apply a new method called sparse Bayesian extreme learning machine (SBELM) for real-time face detection because SBELM can minimize the model size with nearly no compromise on the accuracy and have fast execution time. Several benchmark face datasets were employed for the evaluation of SBELM against other state-of-the-art techniques. Experimental results show that SBELM achieves fastest execution time with high accuracy over the benchmark face datasets. A MATLAB toolbox of SBELM is also available on our Web site.
BibTeX
@article{vong2015fast,
  author = {Chi-Man Vong and Keng Iam Tai and Chi-Man Pun and Pak-Kin Wong},
  title = {{Fast and accurate face detection by sparse Bayesian extreme learning machine}},
  journal = {Neural Computing and Applications},
  year = {2015},
  doi = {10.1007/s00521-014-1803-x},
}

Loading related content...

Figure