The Impact of Color Space in Image Classification Using Machine Learning

A look at how the encoding of colors could improve image clasification

Published on May 30, 2020

Reading time: 0 minutes.


A Mock IEEE Paper for the ‘Brain Computer Interfaces’ course, which covered machine learning. I Designed a simple neural network with TensorFlow, performed various tests to find acceptable hyperparameters using W&B. Then trained identical networks in CIFAR-10 using different color spaces to determine whether “perceptual” or “traditional” color spaces performed better. Due to various factors, results were inconclusive, leaving the door open for further study.

A sample result of my research