Classification and Detection of Cyanosis Images on Lightly and Darkly Pigmented Individual Human Skins using a Fine-Tuned MobileNet Architecture

Abstract

The classification and detection of cyanosis using in-vivo and in-silico image processing approaches are intriguing and very special. In this study, a peripheral and central cyanosis image classification approach, using lightweight-deep learning Convolutional Neural Networks (CNNs), referred to as pre-trained MobileNet architecture, was introduced. This modified MobileNet model was assessed using the sanctioned dataset of 1300-image collected from multiple cyanosis published datasets. The augmentation technique was applied on the training dataset to enrich the productivity. Emphatic results, validation-accuracy and accuracies on the training and test datasets of 95% and 97%, respectively; were obtained as compared to the validation-accuracy of 79% and 82% of the Simple Convolutional Neural Networks (SCNNs) and Fine-tuned VGG16 models attained from prior stud.