Image Based Classification of COVID-19 Infection Using Ensemble of Machine Learning Classifiers and Deep Learning Techniques

Coronavirus illness (COVID-19) had a major impact on multiple areas in society including the healthcare system and the downfall of the global economy. The researchers, doctors, and specialists are working towards new techniques to identify COVID-19 more quickly, such as developing a device that can detect the COVID-19 automatically. In this paper, we propose an automated detection mechanism for identifying COVID-19 patients using a patient’s chest X-ray images. The proposed system made use of CNN (convolutional neural network) and an ensemble of a set of classifiers. The CNN is used for feature extraction in the training and input image whereas classifiers are used for effective prediction. Some of the binary ML (machine learning) classifiers are used for the identification of COVID-19 based on the retrieved characteristics. Later these results are grouped to create a pool of ensemble of classifiers to assure superior results considering various datasets of different sized images with varying resolutions. The performance analysis is discussed and shown as it is better than other previous schemes using deep learning, with 99.17 percent accuracy, 99.19 percent precision, 99.17 percent recall, and 99.43 percent F1 score. The system’s high value in the automated detection of COVID-19 is maintained due to its quick identification and low false-negative rate.