Training convolutional filters for robust face detection

Abstract

We present a face detection approach based on a convolutional neural architecture, designed to detect and precisely localize highly variable face patterns, in complex real world images. Our system automatically synthesizes simple problem-specific feature extractors from a training set of face and non face patterns, without making any assumptions or using any hand-made design concerning the features to extract or the areas of the face pattern to analyze. Experiments on different difficult test sets have shown that our approach provide superior overall detection results, while being computationally more efficient than most of state-of-the-art approaches that require dense scanning and local preprocessing.