A Memory-Efficient Hardware Architecture for Deformable Convolutional Networks | IEEE Conference Publication | IEEE Xplore