Abstract

Convolution has been the superstar method in vision-related tasks. Convolution neural networks, which are networks that apply convolutions in at least one layer, have been some of the most performant in computer vision. There are several advantages of using convolutional layers in a network: dimensionality reduction; translation invariance; transformation invariance; e.t.c. In this blog post, we’re going to explore the ins-and-outs of convolutional neural networks, and try to develop an intuition about why they work so well on vision-related tasks. Furthermore, we’ll some of their frontier applications, and other architectures they’re being used with at the frontier.

Introduction

Convolutional neural networks are extremely powerful. They’ve seen the most success in their application to vision-related tasks, such as object detection, pose estimation, and action detection and recognition.

COMING SOON…