To kick off graphics group this semester, we will start with colors and some color themes. Gretchen Albrecht is a New Zealand painter famous for her use of color. Ben Marwick and Di Cook initiated an R package “gretchenalbrecht” as a way to collect and disseminate color palettes based on Gretchen Albrecht’s paintings. We will investigate and discuss the package with a graphical twist. We will also discuss an avenue to create color schemes automatically.
Read more →
Most of us are quite familiar with the advantages of automating an analysis using R - If we change the data or tweak some parameters, we need only rerun the script in order to produce new results. But one task that we don’t often think of automating is graphics design and image processing. Maybe we need to combine two images side-by-side, or manipulate the dimensions of a set of images to fit some journal convention.
Read more →
A convolutional neural network (CNN) is a tool for deep learning that uses hierarchical networks of hidden layers to learn patterns in a set of training images and detect the presence of similar features in new images. Shoe outsoles are well-suited to classification with CNNs as they often contain distinct and well-defined shapes, such as circles and quadrilaterals, as well as text and textures. Automatic classification of outsole features can be used to identify the make and model of a shoe, speed up database searches, or determine the relative frequency of a particular outsole or set of features within specific populations.
Read more →
Did you know that there are thirteen U.S. federal government agencies tasked with collecting statistics? Can you name any? (Besides the Census Bureau…) In addition to thirteen dedicated agencies, the federal government also has over one hundred other programs dedicated to collecting statistics on various goings-on in the United States. These programs range from the more intuitive, like the Agricultural Research Service (part of the USDA), to the quite niche, like the National Cemetery Administration (part of the VA).
Read more →
Forensic ballistic analysis comprises of identification of whether two bullets come from the same firearm (same-source) or not. This can be thought as an inverse problem to the actual process of firing a bullet, where we have an outcome first, and we then, try to find the source that generated the outcome. Therefore we need a framework of quantifiable steps that takes us from the questioned bullet to a firearm. In this talk I will first give a brief overview of this framework i.
Read more →
