We prepared a new dataset (Cov-Pneum) for X-ray images by processing and merging three well-known publicly available datasets from Kaggle [1]–[4]. This dataset includes a total of 21,272 CXR images of COVID-19 (COVID-19 virus-infected lung), pneumonia (pneumonia-infected lung), and normal (clear lung) and consisting of images 4296, 5824, and 11152, respectively. We applied image scaling and preprocessing operations to enhance the quality of these CXR images.