Abstract.
This paper presents a novel method to solve multi-view face detec- tion problem by Error Correcting Output Codes (ECOC). The motivation is that  face patterns can be divided into separated classes across views, and ECOC  multi-class method can improve the robustness of multi-view face detection  compared with the view-based methods because of its inherent error-tolerant  ability. One key issue with ECOC-based multi-class classifier is how to con- struct effective binary classifiers. Besides applying ECOC to multi-view face  detection, this paper emphasizes on designing efficient binary classifiers by  learning informative features through minimizing the error rate of the ensemble  ECOC multi-class classifier. Aiming at designing efficient binary classifiers, we  employ spatial histograms as the representation, which provide an over- complete set of optional features that can be efficiently computed from the  original images. In addition, the binary classifier is constructed as a coarse to  fine procedure using fast histogram matching followed by accurate Support  Vector Machine (SVM). The experimental results show that the proposed  method is robust to multi-view faces, and achieves performance comparable to  that of state-of-the-art approaches to multi-view face detection.