Table 1.

Properties of the deep feature extractors used in this work and their performance on the ImageNet Challenge.

Feature Extractor	Parameters (M)	Number of Layers	Top-5 Error
AlexNet [19]	61	8	15.3
ZFNet	-	8	14.8
VGG-16 [22]	138	16	7.40
GoogLeNet [23]	6.9	22	6.66
ResNet-50 [24]	25	50	3.57
ResNet-101 [24]	42.6	101	-
ResNetXt-101 [26]	42.6	101	3.03