The Automatic Detection of Pedestrians under the High-Density Conditions by Deep Learning Techniques

Table 3

The statistics of testing sets.

Name	Red ones	Blue ones	Total objects	Total running time (s)	Images chosen	Averaged precision	Averaged recall	Averaged F1 score
16-3-1	42	40	82	106	8	0.986	0.998	0.992
16-4-1	52	52	104	197	14	0.976	0.996	0.986
16-5-1	65	60	125	263	19	0.977	0.994	0.985
16-5-2	64	58	122	250	18	0.985	0.974	0.979
16-6-1	70	76	146	423	29	0.975	0.971	0.973
16-6-2	74	73	147	267	19	0.981	0.978	0.980
16-7-1	78	84	162	555	38	0.981	0.987	0.984
16-7-2	83	83	166	411	28	0.988	0.977	0.982
16-8-1	100	99	199	623	43	0.992	0.959	0.975
16-8-2	99	99	198	360	25	0.987	0.951	0.969
16-9-1	117	117	234	1198	41	0.983	0.958	0.970
16-9-2	111	105	216	383	27	0.990	0.962	0.976
17-3-1	77	68	145	548	38	0.994	0.996	0.995
17-4-1	96	109	205	304	21	0.993	0.986	0.990
17-4-2	100	98	198	205	15	0.994	0.990	0.992
17-5-1	128	120	248	339	24	0.991	0.992	0.991
17-5-2	124	125	249	291	20	0.997	0.985	0.991
17-6-1	148	148	296	362	25	0.989	0.985	0.987
17-6-2	147	154	301	278	20	0.995	0.983	0.989
17-7-1	175	182	357	470	32	0.996	0.984	0.990
17-7-2	176	182	358	357	25	0.997	0.969	0.983
18-3-1	55	51	106	540	37	0.999	1.000	0.999
18-4-1	74	73	147	528	36	0.982	0.995	0.989
18-4-2	74	72	146	295	21	0.981	0.982	0.981
18-5-1	89	86	175	413	29	0.965	0.993	0.979
18-5-2	84	94	178	373	26	0.986	0.989	0.987
18-6-1	110	114	224	428	30	0.980	0.977	0.979
18-6-2	103	106	209	346	24	0.981	0.982	0.982
18-7-1	133	134	267	452	31	0.970	0.977	0.973
18-7-2	126	125	251	456	31	0.987	0.977	0.982
18-8-1	147	147	294	713	49	0.952	0.981	0.965
18-8-2	146	149	295	487	33	0.981	0.971	0.976