Image classification with CNNs

In this section, we will look at some of the most successful CNN architectures for image classification tasks, such as VGGNet, InceptionNet, and ResNet. These networks are also used as feature extractors in object detection models, owing to their great feature extracting capabilities. We will discuss the networks in brief in the following subsection.

VGGNet

VGGNet was developed by K. Simonyan and A. Zisserman, from the University of Oxford. The network was the runner-up at ILSVRC 2014. VGGNet is an improvement over AlexNet, replacing the higher convolution size of 11 and 5 by smaller 3 x 3 convolutions, consistent over multiple stacked layers. Although VGGNet was not the winner of ILSVRC...