Table 13.

Gain/Loss Precision at k for CNN where the best layer is not equal for each k

CNN	Layer	Layer1	P@5	P@10	P@50	P@100
MobileNetV2	block_14_add	block_13_project_BN	-0.0050	-0.00400	0.01255	0.007475
		out_relu	0.0270	0.00850	0.00140	-0.003925
NASNetLarge	Normalizational_concat_12	reduction_concat_reduce_12	0.0295	0.00900	0.00535	-0.000275
VGG16	block5_pool	block4_pool	-0.0130	0.01575	0.03430	0.024600

In bold the best results