Table 4.

Classification results after temporal and categorical pooling. The recall and precision of the different models trained considered the categorical pooling of BEMITA and TRIAVA and the temporal pooling of Lab0d and Lab7d.

ID	Voc size	Colour	Recall in %				Class mean in %
			BEM-TRI	ENCAFO	MACRPY	BKGRND
183	200	Greyscale	94.59	92.37	96.76	99.16	95.72
184	200	RGB	95.00	91.95	97.41	98.88	95.81
185	200	HSV	95.09	92.51	96.44	99.16	95.80
189	500	Greyscale	93.34	92.23	98.38	99.16	95.78
190	500	RGB	93.59	92.37	95.79	99.16	95.23
191	500	HSV	94.59	92.66	95.79	99.16	95.55
			Precision in %
183	200	Greyscale	94.82	99.54	100.00	84.09	94.61
184	200	RGB	95.08	99.54	98.37	85.06	94.51
185	200	HSV	95.25	99.54	99.00	84.89	94.67
189	500	Greyscale	95.00	99.54	98.70	82.13	93.84
190	500	RGB	95.17	99.24	99.00	81.19	93.65
191	500	HSV	95.46	99.39	98.34	83.49	94.17

As previously described, the results for different dictionary sizes (vocsize) and colour spaces (colour) are compared.