Table 2.

Number of retrained cross-validation results for Sheffield and Piracicaba datasets.

Pre-Trained Network	Total Number		Number of
	Layers	Parameters (Millions)	Frozen Layers	Retrained Parameters (Millions)	Image Size
Inception ResNetv2	825	55.9	818	1.5	299 × 299 × 3
Inception-v3	315	23.9	307	2.05	299 × 299 × 3
ResNet-101	347	44.6	79	43.2	224 × 224 × 3
VGG-16	41	138	0	138	224 × 224 × 3