Multimodal Multiobject Tracking by Fusing Deep Appearance Features and Motion Information

Table 3

Ablation study on the training set of KITTI tracking benchmark.

Parameter	MOTA (%)↑	MOTP (%)↑	MT (%)↑	ML (%)↓	IDS↓	FRAG↓
Nt = 0.6	71.27	82.92	43.73	8.42	59	409
Nt = 0.8	70.09	83.11	49.64	6.81	289	559
Fmin = 1	72.05	82.70	46.01	5.02	161	548
Fmin = 5	68.13	83.30	32.97	13.26	35	229
Ours	72.36	83.94	47.31	7.36	94	395