Table 9.

Performance Analysis of the proposed work with current research works that have utilized the same chest X-ray data sources for COVID-19, Pneumonia images as the current work.

S No	Source	Methodology	Class	Number of COVID-19 Test samples	Approx. parameters	Overall F1-score (%)	Overall accuracy (%)	F1-score for COVID-19 (%)	COVID-19 class accuracy (%)
1	Ozturk et al. [39]	DarkNet-19 based CNN	3	$\sim$25	1.164M	87.40	87.02	88.00	87.02
2	Mangal et al. [34]	CheXNet based CNN	4	30	26M	92.30	87.2	96.77	99.6
3	Khan et al. [33]	Transfer learning with Xception net	4	$\sim$70	33M	89.8	89.6	95.61	96.6
4	Wang and Wong [42]	Customized CNN architecture	3	100	11.75M	93.13	93.33	94.78	96.67
5	Apostolopoulos and Mpesiana [31]	Transfer learning with MobileNetV2	4	222	3.4M	93.80	94.72	90.50	96.80
6	Farooq and Hafeez [30]	ResNet50 based CNN	4	8	25.6M	96.88	96.23	100.0	100.0
7	Proposed Work	Customized CNN with distinctive filter learning module	4	112	15.6M	96.90	97.94	97.20	99.80