A Face Detection Framework Based on Deep Cascaded Full Convolutional Neural Networks

In practical applications, the collected face images are often affected by the surrounding environment, resulting in multiple expressions, multiple poses, occlusion, light intensity, and complex background issues in face detection. Hence this research presents a novel face detection framework based on deep cascaded full convolutional neural networks (CNNs) to solve the mentioned issues in face detection. This frame work also supports face detection, and positioning of face key points at the same time by using its 3-phase cascaded CNN architecture. The 3-phase cascaded architecture is the combination of three phases of three CNN layers; phase1 (I-Net, Initial Network), phase2 (A-Net, Advanced Network), and phase3 (U-Net, Ultimate Network). In this CNN design, Inception Architecture (spilt-merge), and a convolutional layer with stride of 2 are used instead of the pooling layer. A bottleneck structure of size 1 x 1 convolution kernel is used for convolution layers, which replaces large convolution kernels with multiple small convolution kernels. Hence the whole network uses a convolutional layer instead of the fully connected layer (FC). The carried-out experiments proved that the presented framework is an effective one for face detection applications.