ColoRectalCADx: Expeditious Recognition of Colorectal Cancer with Integrated Convolutional Neural Networks and Visual Explanations Using Mixed Dataset Evidence

Table 3

Number of CNN parameters of CNNs.

CNN architecture models	Introduced year	Total params	Trainable params	Nontrainable params	Layers
AlexNet [35, 36]	2012	2,81,02,775	2,80,81,639	21,136	23
DarkNet-19 [37]	2017	1,60,45,847	1,60,32,983	12,864	19
ResNet-50v2 [38]	2016	2,59,33,975	23,69,175	2,35,64,800	50
DenseNet-201 [39] [40]	2018	1,94,29,463	11,07,479	1,83,21,984	201
EfficientNet-B7 [41]	2019	6,55,73,799	14,76,119	6,40,97,680	813
VGG-16 [42]	2014	1,53,14,391	5,99,703	1,47,14,688	16
VGG-19 [43]	2014	2,06,24,087	5,99,703	2,00,24,384	13
NasNetLarge [44]	2018	87,256,682	2,339,864	84,916,818	414
InceptionResNetV2 [45]	2016	55,398,648	1,061,912	54,336,736	164
Proposed integrated models
ADaRDEV^2I-22	2022	246,429,456	71,359,568	175,069,888
ADaRDEV^2-22		19,10,28,063	7,02,94,911	12,07,33,152
RDEV^2-22		14,68,78,383	2,61,79,231	12,06,99,152
ADaDR-22		8,94,87,664	4,75,66,880	4,19,20,784
ADaR-22		2,59,33,000	23,68,200	2,35,64,800
DaRD-22		6,13,99,840	1,95,00,192	4,18,99,648
DEV-22		100,319,600	3,185,248	97,134,352
ADa-22		4,41,26,048	4,40,92,048	34,000
RV-22		46,559,368	22,994,568	23,564,800
RD-22		4,53,61,624	34,74,840	4,18,86,784
AD-22		4,75,16,384	2,91,73,264	1,83,43,120
DaR-22		4,19,67,694	1,83,90,030	2,35,77,664