Thai Finger-Spelling Recognition Using a Cascaded Classifier Based on Histogram of Orientation Gradient Features

Table 3

The performance of 21 trained classifiers (adjusting ).

Class	(2nd experiment)
	Hit	Miss	False	Hit	Miss	False	Hit	Miss	False	Hit	Miss	False	Hit	Miss	False	Hit	Miss	False
“1”	46	4	247	41	9	301	44	6	127	38	12	44	20	30	7	9	21	5
“2”	49	1	129	48	2	32	45	5	9	45	5	2	21	29	0	31	19	0
“3”	49	1	32	48	2	8	44	6	5	44	6	3	45	5	3	43	7	5
“4”	41	9	24	33	17	14	27	23	7	25	25	0	18	32	3	12	38	0
“5”	48	2	91	46	4	29	42	8	8	36	14	5	32	18	1	35	15	1
“”	50	0	69	50	0	41	50	0	8	50	0	7	49	1	1	49	1	1
“”	42	8	276	31	19	99	13	37	29	7	43	9	3	47	2	2	48	0
“”	48	2	42	42	8	13	43	7	8	37	13	6	43	7	0	36	14	1
“”	44	6	164	41	9	28	40	10	12	30	20	2	31	19	0	22	28	1
“”	33	17	376	19	31	141	21	29	37	11	39	7	7	43	10	9	41	4
“”	45	5	95	40	10	50	44	6	7	40	10	2	36	14	0	37	13	2
“”	47	3	21	39	11	8	41	9	0	24	26	0	31	19	0	26	24	0
“”	47	3	0	44	6	0	42	8	0	47	3	0	41	9	0	45	5	0
“”	43	7	194	42	8	148	44	6	76	34	16	14	23	27	8	20	30	9
“”	45	5	159	43	7	69	42	8	46	33	17	6	31	19	12	24	26	0
“”	46	4	153	44	6	102	42	8	26	36	14	11	35	15	0	36	14	1
“”	46	4	83	48	2	52	41	9	10	36	14	3	22	28	2	18	32	1
“”	50	0	22	49	1	16	42	8	0	43	7	1	36	14	1	34	16	1
“”	47	3	42	44	6	18	42	8	4	33	17	2	34	16	0	26	24	0
“”	40	10	278	40	10	252	38	12	141	38	12	17	16	34	7	11	39	3
“”	46	4	39	44	6	65	44	6	11	42	8	0	41	9	0	39	11	1