Neuromorphic Vision Based Multivehicle Detection and Tracking for Intelligent Transportation System

Table 5

The tracking performance for EventSeq-Vehicle2 using different tracking methods (SORT, GM-PHD, GM-CPHD, and PDAF) with detection through DBSCAN clustering. The impact of time intervals from 10ms to 30ms was studied.

Tis	Tracker	MOTA	MOTP	MT	PT	ML	FP	FN	IDs	FM
10ms	SORT	24.4%	70.2%	3	53	29	2170	14929	183	942
10ms	GM-PHD	13.4%	69.4%	0	60	25	3807	14994	1000	2524
10ms	GM-CPHD	7.8%	69.7%	2	69	14	7452	13084	528	2524
10ms	PDAF	13.8%	69.8%	0	57	28	4403	15206	100	3050
20ms	SORT	5.7%	68.1%	7	49	28	3304	7393	81	331
20ms	GM-PHD	15.6%	70.6%	11	52	21	2839	6524	290	729
20ms	GM-CPHD	11.3%	70.6%	10	57	17	3824	6196	118	655
20ms	PDAF	11.5%	70.5%	4	58	22	3091	6948	76	995
30ms	SORT	0%	67.3%	3	45	37	2069	5496	53	183
30ms	GM-PHD	7.6%	70.3%	5	50	30	1888	5001	149	328
30ms	GM-CPHD	-0.7%	70.1%	4	57	24	2872	4694	100	389
30ms	PDAF	5%	69.9%	3	55	27	1969	5216	49	542