Face localization is a face detection problem where the number of people is known. We present a comparison between different algorithms fusion methods dedicated to the localization of faces in color images. Data to combine result from an appearance model supported by an auto-associative network, an ellipse model based on Generalized Hough Transform, and a skin color model. We intro-duce and compare several fusion methods like the Bayesian classifier with parametric or non-parametric technique, a fuzzy inference system, and a weighted average. Given an input image, we compute a kind of probability map on it using a sliding window. The face position is then determined as the location of the absolute maximum over this map. Improvement of basic detectors localization rates is clearly shown and prevalence of the weighted average is reported.