Detection of Covid-19 from Chest X-Ray Images

The capability of deep learning on chest radiographs is diagnosed in this dissertation, and an image classifier based on the COVID-Net has been provided to categorize chest X-Ray pictures. Information augmentation was recommended to extend COVID-19 reports 17 times in the case of a little amount of COVID-19 metadata. The system aims to improve transfer learning and model integration by categorizing chest X-ray pictures into three categories: normal, COVID-19, and infectious pneumonia. Choose the models ResNet-101 and ResNet-152 with best outcomes for fusion based on accuracy and loss value, and apathetically enhance its weight percentage during the training procedure. The system can obtain 97 percent accuracy in chest X-Ray pictures post training. In the diagnosing of lung nodules, this technique offers a higher accuracy than radiologist. It can assist radiologists enhance their efficiency levels and diagnostic performance as an additional diagnostic technique. Keywords: COVID-19, pneumonia, x-ray, convolutional neural networks, Coronavirus.