A Multilevel Single Stage Network for Face Detection

Figure 2

Enhanced feature module. The enhanced feature module is composed of two common convolutions and two dilated convolutions.