Comparative Study of the Effect of Various Color Models on Classification of Parasitic Eggs

Parasitic infections are now recognized by the World Health Organization (WHO) as one of the main causes of illness. Automating routine feces inspection for parasitic illnesses is the goal of this challenge, which aims to bring together experts in the field to develop reliable automated algorithms to detect and categorize parasitic worm eggs in a range of microscopic images. In this study, we evaluated how various color models affected the classification of 11 different types of parasite eggs from fecal smear samples. The input of the classification algorithm is the preprocessed microscopic image. Seven color spaces classifiers have been tested on the Inception-ResN et-V2 CNN model. Additionally, an analysis of variance (ANOVA) has been conducted to compare the expected error rate of those seven classifiers. The models were tested on 2,200 entirely separated images after being trained on 11,000 images. Out of seven color spaces on the test dataset, the XYZ color space performed better (accuracy: 82.77%, recall: 82.77%, precision: 84.53%, and Fl:83.64%). According to the results of the ANOVA analysis, the error rates produced by those seven classifiers are not statistically different.