Predicting Artist Drawing Activity via Multi-Camera Inputs for Co-Creative Drawing

Jansen, Chipp and Sklar, Elizabeth (2021) Predicting Artist Drawing Activity via Multi-Camera Inputs for Co-Creative Drawing. Proceedings of the 22nd Towards Autonomous Robotic Systems (TAROS) Conference . ISSN TBD

Documents
Predicting Artist Drawing Activity via Multi-Camera Inputs for Co-Creative Drawing
[img] PDF
AWARD-TAROS2021paper64.pdf - Whole Document
Restricted to Repository staff only

668kB
Item Type:Article
Item Status:Live Archive

Abstract

This paper presents the results of experimentation in computer vision based for the perception of the artist drawing with analog media (pen and paper), with the aim to contribute towards a human- robot co-creative drawing framework. Using data gathered from user studies with artists and illustrators, two types of CNN models were de- signed and evaluated to predict an artist’s activity (e.g. are they drawing or not?) and the position of the pen on the canvas based only on a multi- camera input of the drawing surface. Results of different combination of input sources are presented, with an overall mean accuracy of 95% (std: 7%) for predicting when the artist is present and 68% (std: 15%) for predicting when the artist is drawing; and mean squared normalised error of 0.0034 (std: 0.0099) of predicting the pen’s position on the drawing canvas. These results point toward an autonomous robotic system having an awareness of an artist at work via camera based input and contributes toward the development of a more fluid physical to digital workflow for creative content creation.

Keywords:human-robot collaboration, co-creative drawing, computer vision, deep learning, sketch-based computing
Subjects:G Mathematical and Computer Sciences > G700 Artificial Intelligence
G Mathematical and Computer Sciences > G400 Computer Science
G Mathematical and Computer Sciences > G760 Machine Learning
Divisions:College of Science > Lincoln Institute for Agri-Food Technology
ID Code:46480
Deposited On:29 Sep 2021 09:50

Repository Staff Only: item control page