ABSTRACT
The world has witnessed one of the most devastating phases in the history of mankind after being hit with the COVID-19 pandemic which still continues to spread rapidly all across the globe. The disease is believed to majorly cause respiratory disorders in humans. Detecting COVID-19 patients through X-Ray images is the only way to slow down the expansion of the pandemic, detecting pneumonia has equally become a demanding task as both exhibit similar properties of affecting the human lungs. Pneumonia is said to be an illness caused by a bacteria in the alveoli of lungs that may accompany to the death of an individual if its treatment is ignored. Hence, developing an automated system to detect the disease can be beneficial to the human race. With continuous progressions in the expertise of deep learning and machine learning;its fundamentals are observed to continuously contribute towards analysis of medical images and classification of patients exhibiting the disease. In this work, we appraise the concepts of ResNet50v2 model and capsule network to predict the affected and unaffected patients using chest X-Ray images. The authors propose a novel classification framework consisting of a convolutional layer, primary capsule layer and digit capsule layer, wherein the radiographic images are categorized through dynamic routing followed by disease prediction through ResNet50v2 model. The proposed work is implemented on images with a resolution of $224 \times 224$ and a batch size of 10. Further, parametric functions are applied to verify the model being trained. © 2022 IEEE.