Convolutional face finder: a neural architecture for fast and robust face detection

Abstract

In this paper, we present a novel face detection approach based on a convolutional neural architecture, designed to robustly detect highly variable face patterns, rotated up to +/-20 degrees in image plane and turned up to +/-60 degrees, in complex real world images. The proposed system automatically synthesizes simple problem-specific feature extractors from a training set of face and nonface 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. The face detection procedure acts like a pipeline of simple convolution and subsampling modules that treat the raw input image as a whole. We therefore show that an efficient face detection system does not require any costly local preprocessing before classification of image areas. The proposed scheme provides very high detection rate with a particularly low level of false positives, demonstrated on difficult test sets, without requiring the use of multiple networks for handling difficult cases. We present extensive experimental results illustrating the efficiency of the proposed approach on difficult test sets and including an in-depth sensitivity analysis with respect to the degrees of variability of the face patterns.

Dealing with Inaccurate Face Detection for Automatic Gender Recognition with Partially Occluded Faces

“…Most of them dealt with face recognition [1,2] and face detection [3][4][5][6] problems. However, automatic gender classification has recently become an important issue in this area.…”

A survey on face detection in the wild: Past, present and future

“…Multi-view face detection was achieved by measuring the reconstruction errors of multiple CGMs, combined via a conditional mixture and an MLP gate network. In [145], the authors proposed a face detection scheme based on a convolutional neural architecture. Compared with traditional feature-based approaches, convolutional neural network derives problem-specific feature extractors from the training examples automatically, without making any assumptions about the features to extract or the areas of the face patterns to analyse.…”

Sign Language Recognition Using Motion History Volume and Hybrid Neural Networks

“…For the pattern classification, we have adopted the weighted FMM model [9] which can provide a feature analysis facility using a feature relevance measure. Convolutional neural networks (CNN) incorporate constraints and achieve some degree of shift and deformation invariance using spatial subsampling and local receptive fields [6]. When an image pattern is input, spatially-localized subset of units (receptive fields) are passed through the two-dimensional processing element in the subsequent layers.…”

“…Yilmaz et al proposed a novel action representation method named action sketch which is generated from a view-invariant action volume [5]. Convolutional neural networks (CNN) have been successfully applied to object recognition in 2D images [6]. The CNN model is a biologically inspired hierarchical multilayered structure, where each sub-layers incorporate feature extraction and feature reduction.…”