Detroit Colour Council presents:
Detroit Colour Council Virtual Meeting
Dimitris Mylonas, “Augmenting Color Communication in Context”
Alexandros Koliousis, “Multi-modal Color-naming Systems with Deep Learning”
Date: Tuesday, May 24, 2022
Time: 1:00pm EST
To Register, click here
Dimitris is an Assistant Professor in Computer Science at New College of the Humanities at Northeastern. Dimitris gained his PhD in colour communication from the Department of Computer Science, University College London under the supervision of Prof. Lewis Griffin, Prof. Andrew Stockman and Prof. Lindsay MacDonald. He obtained a MSc in Digital Colour Imaging from London College of Communication, University of the Arts London under the supervision of Prof. Lindsay MacDonald and completed successfully a MRes in Media and Arts Technology at Queen Mary University of London under the supervision of Prof. Matthew Purver and Prof. Jules Davidoff. He held research posts in the School of Psychology, University of Liverpool under the supervision of Prof. Sophie Wuerger, in the Wellcome Laboratory of Neurobiology, University College London under the supervision of Prof. Semir Zeki and in the Dept. Psychology, Goldsmiths College, University of London under the supervision of Prof. Jules Davidoff. Dimitri’s interdisciplinary research on colour cognition has been widely published and received awards and recognitions. Dimitris is Chairman of the study group ‘Language of Colour’ of the International Colour Association (AIC) and Committee member of the Colour Group (GB).
Augmenting Color Communication in Context
Color naming links vision and speech. It refers to our intriguing cognitive capacity to group millions of discriminable colors into a smaller set of categories named, for example, “yellow”, “teal” or “periwinkle blue”. Since 2009, we have led an international collaborative project to collect unconstrained colour names with their corresponding colour ranges through a crowdsourcing experiment with thousands of observers in tens of languages (accessible at https://colornaming.com) and compared the results against psychophysical methods performed in controlled laboratory viewing conditions. We developed robust computational tools trained by these multilingual datasets to automate the color naming task across the full color gamut and we are now expanding our data collection and computational methods to automate the color naming task in complex scenes. Through discussion with key companies in the industry, we have identified specific gaps in current capability with immediate applicability for our tools.
Dimitris Mylonas https://www.linkedin.com/in/dimitrismylonas/?originalSubdomain=uk
Alexandros Koliousis is an Associate Professor of Computer Science at the New College of the Humanities and member of the Institute for Experiential AI at Northeastern University. His current research interests lie at the intersection of scalable data systems and deep learning.
Koliousis has worked on the design and implementation of high-performance data-parallel multi-GPU processing systems in the areas of deep learning and data stream processing. He has also researched topics including efficient natural language processing in hardware, complex event processing for home network management, and routing systems for wireless sensor networks. Before joining Northeastern’s London affiliate, Koliousis held an industry research position at the semiconductor company Graphcore and academic positions at the Imperial College London and the University of Glasgow. He earned his doctoral degree in computing science and his Master of Science in advanced computing science from the School of Computing Science at the University of Glasgow.
Multi-modal Color-naming Systems with Deep Learning
Color is fundamental in our perception from object recognition to visual memory. We study cognitive processes from color vision to communication. Armed with experimental data that map colors to words, part gathered in controlled laboratory conditions and part in a web-based setting, we aim to evaluate the color naming accuracy of deep learning vision-language models in complex scenes and across languages.
Alexandros Koliousis https://www.linkedin.com/in/alexandros-koliousis/
Membership is free at: https://detroitcc.org/join-the-dcc
The Detroit Colour Council is a professional society based in the metropolitan Detroit, Michigan area, which promotes communication and education around color and design quality issues with emphasis in the automotive sector. Members represent a wide variety of professional functions including engineers, designers, salespeople, scientists, consultants, lab technicians and students.