Fig. 4From: Efficient 3D object recognition in mobile edge environmentThe architecture of our multi-view CNN modelBack to article page