Many countries are severely affected by COVID-19, and massive efforts are required to mitigate the terrible impacts of the COVID-19 global epidemic. To stop COVID-19 from spreading widely and minimize COVID-19 patient treatment issues, early discovery of the virus in its early stages is essential. Conventional methods for COVID-19 detection are time-consuming, cumbersome, unreliable, and yield an incorrect diagnosis. Most deep learning (DL) methods for COVID-19 identification described in the past only support one data type (chest radiographs or CT scan images). Examining unusual patterns in various data types may yield high classification performance. For this purpose, we use two image datasets: chest radiographs and chest computer tomography (CT). This research study proposed an excellent COVID-19 detection strategy based on the Resnet-18 deep convolutional neural network (CNN) to timely and efficiently identify COVID-19 spread. Also, we used each image data type to conduct extensive Covid-19 detection experiments with the proposed DL model. We performed a trinary classification task (normal, covid-19, and pneumonia) using chest radiographs and a binary classification (COVID-19 and Normal) using CT scan images. We validated the performance of the proposed approach on two standard Kaggle datasets, i.e., the COVID-19 Radiography Database and the SARS-CoV-2 CT scan dataset. The Resnet model's detection accuracy of 99.42% using chest radiographs (CRs) and 100% using CT scans for COVID-19 detection indicates its efficacy in detecting COVID-19.