A big “Thank You!” to all that came to say “Hi” at Mobile HCI-2020 and asked questions about our work.
Abstract
This poster was presented at Mobile HCI 2020. In this work we present a way to calculate optimal transformation parameters for the Crayon approximate colour transform. We show that when using the least squares cost, the relationship between the user score and the most aggressive transformation parameter that can achieve that score follows an exponential shape. We use that to learn (simple) models that can predict the shape of this exponential using simple image colour based heuristics. We show that our models can predict this shape down to an error of 1.8% and that its robust against outliers.
