TABLE 10. Comparison of Models Performance on Imbalanced Datasets.

Reference	Model	Datasize	Data ratio	Accuracy
Brunese et al. [42]	VGG16	6505	3003: 3520 ≈ 1: 1.17	99%
Apostolopoulos & Mpesiana [36]	CNN	1428	224: 1208 ≈ 1: 5.4	98%
Ozturk et al. [37]	DarkNet	1750	250: 1500=1: 6	98%
Khan et al. [38]	Xception	1251	284: 967 ≈ 1: 3.4	89.6%
Lee et al. [67]	VGG-16	1821	607: 1214=1: 2	95.9%
Shelke et al. [68]	DenseNet-VGG16	2271	500: 1771 ≈ 1: 3.54	95%
Best model from Study Two with 95% CI	VGG 16 InceptionResNetV2 MobileNetV2 VGG 19	1845	262: 1583≈1:6.04	97.6%–99.7%
				97.6%–99.7%
				97.6%–99.7%
				95.2%–98.6%