This repository contains the files to various CNN architectures that have been trained with the popular Kaggle Chest X-Ray dataset to determine whether a medical photograph of the lungs of a patient contain pneumonia, or are healthy.
The model that performed best for this classification task (pneumonia or healthy) was ResNeXt-152 (~92.9% accuracy). The model that performed the worst was EfficientNet-B0 (~59.3% accuracy).