Table 5. Classification performance of ensembles and CNN learning machines in the test set.

The best ensemble and single view classifiers for each performance metric are highlighted in bold and blue, respectively. H = head view; D = dorsal view; P = profile view; G = general classifier; S = specific classifier; T = transfer classifier; E = ensemble classifier.

		G			S			T			E
		H	D	P	H	D	P	H	D	P	G	S	T	All
Accuracy	Top 1	0.77	0.61	0.64	0.76	0.54	0.60	$0.78$	0.59	0.65	0.81	0.79	0.81	0.83
	Top 3	$0.91$	0.80	0.82	0.90	0.74	0.78	$0.91$	0.78	0.83	0.92	0.90	0.92	0.93
	Top 5	$0.94$	0.87	0.88	$0.94$	0.82	0.85	$0.94$	0.85	0.89	0.95	0.94	0.95	0.96
Average precision	Top 1	$0.74$	0.53	0.54	0.72	0.46	0.51	$0.74$	0.52	0.57	0.80	0.78	0.79	0.83
	Top 3	$0.91$	0.79	0.81	0.89	0.74	0.77	$0.91$	0.77	0.81	0.92	0.91	0.92	0.94
	Top 5	0.94	0.86	0.87	0.94	0.83	0.84	$0.95$	0.85	0.89	0.95	0.95	0.96	0.96
Minimum precision	Top 1	$0.32$	0.11	0.24	0.31	0.00	0.20	0.30	0.16	0.24	0.45	0.39	0.43	0.40
	Top 3	0.76	0.54	0.58	$0.79$	0.56	0.60	$0.79$	0.53	0.64	0.81	0.81	0.83	0.84
	Top 5	0.86	0.74	0.74	0.86	0.70	0.73	$0.89$	0.69	0.81	0.89	0.91	0.92	0.90