Convolution Can Incur Foveation Effects

Jun Yuan, Bilal Alsallakh, Narine Kokhlikyan, Vivek Miglani, Orion Reblitz-Richardson

Keywords: CNN, convolution, foveation, padding, boundary treatment, visualization

TL;DR: An interactive visualization to illustrate potential foveation effects incurred during convolution

Abstract: This exhibit demonstrates how boundary treatment in convolutional networks can incur foveation effects: Impacted pixels have fewer ways to contribute to the computation than central pixels. Different padding mechanisms can either eliminate or aggravate these effects, which is made obvious by an interactive visualization.

Exhibit: https://mind-the-pad.github.io/