A family of super deep networks, referred to as residual networks or ResNet,
achieved record-beating performance in various visual tasks such as image
recognition, object detection, and semantic segmentation. The ability to train
very deep networks naturally pushed the researchers to u