Table 4. Optimised features and model hyperparameters.

a: Random forest classifier.
Feature	Variable	MHNGr.ET.1023 (dog)	MHNGr.ET.1017 (raptor)	MHNGr.ET.1456 (ibis)	MG.2038 (ibis in a jar)
Histogram 1	Radius	3	6	5	7
	Bins	32	8	8	32
Histogram 2	Scale	0	0	0	2
	Radius	12	27	21	24
	Bins	16	32	32	16
LBP	Scale	0	1	1	0
	Radius	15	27	20	30
Random forest	Number of trees	64	64	64	64
b: Neural network classifier.
Feature	Variable	MHNGr.ET.1023 (dog)	MHNGr.ET.1017 (raptor)	MHNGr.ET.1456 (ibis)	MG.2038 (ibis in a jar)
Histogram 1	Radius	7	8	7	7
	Bins	16	32	8	32
Histogram 2	Scale	1	2	0	2
	Radius	6	32	27	29
	Bins	8	32	32	16
LBP	Scale	1	1	1	0
	Radius	18	28	27	28
Neural network	Layer 1 size	256	256	64	256
	Layer 2 size	64	64	128	64
	Dropout rate	0.4	0.45	0.25	0.25
	Init. stddev.	0.0001	0.0251	0.1585	0.0016
	Minibatch size	32	32	32	32