A compact CNN model for automated detection of COVID-19 using thorax x-ray images

COVID-19 is an epidemic, causing enormous death toll. The mutational changing of RNA virus diagnostic complexities. RT-PCR and Rapid Tests are used for the diagnosis, but unfortunately, these methods ineffective in diagnosing all strains COVID-19. There utmost need to develop a procedure timely identification. In proposed work, we come up with lightweight algorithm based on deep learning rapid detection system thorax chest x-ray (CXR) images. This research aims fine-tuned convolutional neural network (CNN) model using improved EfficientNetB5. Design compound scaling trained best possible feature extraction algorithm. low convergence rate work can be easily deployed into limited computational resources. It will helpful triaging victims. 2-fold cross-validation further improves performance. trained, validated, testing performed form internal external validation self-collected compiled real-time dataset CXR. training relatively extensive compared existing ones. performance technique measured, other state-of-the-art pre-trained models. methodology gives remarkable accuracy (99.5%) recall biclassification. two different test also give exceptional predictions. visual depiction predictions represented by Grad-CAM maps, presenting extracted features predicted results.