Automated Diagnosis of Pneumonia from Classification of Chest X-Ray Im ages using EfficientNet

Pneumonia is a fatal contagious agent that causes respiratory disorders. The methodology utilized by an advisor to evaluate pneumonia through chest X-ray images is tedious and pricey. However, it also requires an experienced and skilled radiologist to correctly interpret the chest x-ray images. Moreover, the diagnosis can be confusing some times. Subsequently, researchers should pay attention to deploying a PC supported framework to recognize pneumonia. Again, clinical and experimental data can come from many different laboratories following different protocols and so databases are usually particularly disordered and noisy. To solve this issue, We have utilized EDA (Exploratory Data Analysis) to interpret the data set to clarify its basic traits. The class imbalance issue is reduced and data augmentation is also applied. A recently designed CNN (convolutional neural network) model EfficientNet is used to train the classifier by transfer learning and fine-tuning. The model has achieved a test accuracy of 96.33% and an AUC(Area Under Curve) of 0.991. We have also calculated other strategic objectives, such as precision, recall, and F1 score, to demonstrate the efficacy of our approach in finding pneumonia in the exclusion of trained physicians.