In this paper, a method of combining face detectors is proposed, which is based on the geometry of the competing face detection results. The main idea of the method consists in finding groups of similar face detection results obtained by several algorithms and further averaging them. The combination result essentially depends on the number of algorithms that have fallen in each of the groups. The experimental evaluation of the method is based on seven algorithms: Viola-Jones (OpenCV 1.0), Luxand© FaceSDK, Face Detection Library, SIFinder, Algorithm of the University of Surrey, FaceOnIt, Neurotechnology© VeriLook. The paper contains practical results of their combination and a discussion of future improvements. © 2011 Springer-Verlag.
CITATION STYLE
Degtyarev, N., & Seredin, O. (2011). A geometric approach to face detector combining. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6713 LNCS, pp. 299–308). https://doi.org/10.1007/978-3-642-21557-5_32
Mendeley helps you to discover research relevant for your work.