Geforce gtx1060 3gb graphics card

In this research, we have constructed training and testing datasets from the award winning Kaggle CXR public database (Chowdhury et al., 2020 ), comprising 219 COVID-19 positive images, 1345 viral pneumonia images and 1341 normal images. Originally these images are of dimension 1024 × 1024 in png format. The database is divided into two distinct training and testing subsets, with images resized to 227 × 227. Then the training dataset is augmented with images generated by applying the translation, rotation and scaling operations on the training images. The description of the dataset is given in Table 1. The proposed framework is implemented with Matlab 2021b software in an i7-7700K processor with 32GB DDR4 RAM equipped with a NVIDIA GeForce GTX1060 3GB Graphics card.
Table 1

Description of training and testing dataset.

Infection	No. of images in database	No. of training images	No. of training images after image augmentation	No. of testing images
				Binary classification	Multiclass classification
COVID-19	219	119	476	100	100
Viral Pneumonia	1345	645	2580	1400	700
Normal	1341	641	2564		700