Table 2.

Training and validation accuracy/loss, precision, recall, and F1-score along with the number of parameters, training time, and epochs required to train deep learning architectures (in the order of the lowest to the highest F1-score).

Deep Learning Architectures	Parameters (in Millions)	Epochs Required to Train the Model	Training Time (in Hours)	Training Accuracy	Validation Accuracy	Training Loss	Validation Loss	Precision	Recall	F1-score
LeafNet	0.324 M	59	5.95	0.8590	0.7961	0.4563	0.6658	0.7946	0.7971	0.7958
VGG-16	138 M	59	38.13	0.8339	0.8189	0.5328	0.5651	0.8182	0.8194	0.8188
OverFeat	141.8 M	58	6.75	0.8995	0.8603	0.3201	0.4330	0.8592	0.8628	0.8610
Improved Cifar-10	2.43 M	58	6.08	0.9256	0.8974	0.2628	0.3205	0.8944	0.8960	0.8952
Inception ResNet v2	54.3 M	58	32.83	0.9551	0.9091	0.1530	0.3047	0.9075	0.9105	0.9089
Reduced MobileNet	0.5 M	55	11.72	0.9570	0.9278	0.1860	0.2442	0.9269	0.9267	0.9268
Modified MobileNet	0.5 M	53	6.38	0.9534	0.9297	0.1632	0.2385	0.9278	0.9265	0.9271
ResNet-50	23.6 M	55	26.33	0.9873	0.9423	0.0468	0.1923	0.9351	0.9358	0.9354
MLCNN	78 M	57	67.33	0.9583	0.9402	0.1335	0.1820	0.9386	0.9411	0.9398
Inception v4	41.2 M	59	52.92	0.9586	0.9489	0.1410	0.1828	0.9410	0.9466	0.9438
Improved GoogLeNet	6.8 M	53	9.67	0.9829	0.9521	0.0522	0.1038	0.9528	0.9539	0.9533
AlexNet	60 M	54	6.10	0.9689	0.9578	0.1046	0.1298	0.9563	0.9570	0.9566
DenseNet-121	7.1 M	56	28.75	0.9826	0.9580	0.0758	0.1323	0.9581	0.9569	0.9575
MobileNet	3.2 M	47	14.70	0.9764	0.9632	0.0903	0.1090	0.9624	0.9612	0.9618
Hybrid AlexNet with VGG (AgroAVNET)	238 M	54	49.90	0.9841	0.9649	0.0546	0.1078	0.9626	0.9674	0.9650
ZFNet	58.5 M	47	6.47	0.9752	0.9717	0.0746	0.1139	0.9746	0.9751	0.9748
Cascaded AlexNet and GoogLeNet	5.6 M	57	6.5	0.9931	0.9818	0.0229	0.0592	0.9749	0.9751	0.9750
Xception	22.8 M	34	56.28	0.9990	0.9798	0.0140	0.0621	0.9764	0.9767	0.9765