October 29, Tuesday, 08:00 – 10:00 (2 hours)
Instructor: Nathan Moroney, consultant
Level: Introductory
Prerequisites: Python is used, but not required. A basic understanding of color science is helpful, but not required.
Benefits This course enables the attendee to:
• Gain a better understanding of machine learning classifiers via color naming.
• Gain insights about color science via machine learning.
• Acquire hands-on experience with a range of algorithms.
• Explore applications of machine color naming to digital imaging.
• Be exposed to benchmarking metrics and visualization tools.
• Obtain Python code samples and relevant datasets.
Course Description
From isolation forests to large language models, color naming is considered in the context of computational color categorization. Machine learning includes a wide range of classification algorithms from logistic regression to support vector machines to k-nearest neighbors and more. This course teaches these algorithms and additional techniques using color as the common feature across all of the methods. In addition, applications to digital imaging such as image stylization, analysis, and similarity measures are included. Python code is used to provide detailed, working examples to get hands-on experience with the data and models. This also includes a range of visualization techniques to better understand the differences between the models.
Intended Audience: engineers, students, and researchers wanting to learn and explore topics at the intersection of color science, machine learning, and digital imaging.
Nathan Moroney has published 90 papers and patents with more than 1,000 combined citations for contributions in color science, digital imaging, and geometry processing. He was the technical chair for CIE TC 8-01 which developed the CIECAM02 color appearance model and he learned the color term ‘chartreuse’ as a result of his first implementation of a machine color naming algorithm. He is a graduate of the RIT Munsell Color Science Laboratory and this course is based on the draft of a manuscript being written on the same topic.