After the deep learning breakthrough in 2012, research toward more refined classification systems based on convolutional neural networks (CNNs) gained momentum. Innovation is moving at a frantic pace nowadays, as more and more companies are developing smart products. Among the numerous solutions developed over the years for object classification, some have became famous for their contributions to computer vision. They have been derived and adapted for so many different applications that they have achieved must-know status, and so deserve their own chapter.
In parallel with the advanced network architectures introduced by these solutions, other methods have been explored to better prepare CNNs for their specific ...
