ABSTRACT
Ever since an anonymous disease broke out in late 2019, the whole world seems to have own ceased functioning. COVID-19 patients are proliferating at an exponential rate, straining healthcare systems around the world. Traditional techniques of screening every patient with a respiratory disease is unfeasible due to the restricted number of testing kits available. We presented a method for recognizing COVID-19 infected patients utilizing data collected from chest X-ray scans to overcome this challenge. This attempt will benefit both patients and doctors significantly. It becomes even more critical in nations where the number of people affected far outnumbers the number of laboratory kits available to test the disease. When current systems are confused whether to retain the patient on the ward with other patients or isolate them in COVID-19 zones, this could be useful in an inpatient setting. Apart from that, it would aid in the identification of patients with a high risk of COVID-19 and a false negative RT-PCR who would require a repeat. Most of the COVID-19 detection methods use traditional image classification models. This has the issue of low detection accuracy and incorrect COVID-19 detection. This method starts with a chest x-ray enhancement procedure like this: Rotation, translation, random conversion. The survey's accuracy has considerably increased as a result of this. For the COVID-19 infection, our model has 97.5 percent accuracy and 100 percent sensitivity (recall). In addition, we used a visualization technique that distinguishes our model from the others by displaying contaminated areas in X-ray pictures. © 2022 IEEE.