Multimodal Multiobject Tracking by Fusing Deep Appearance Features and Motion Information

Figure 8

Result of the qualitative evaluation with the test sequence 1. (a) The detector generates an error detection and disappears in the next frame. (b) The proposed MOT method can track objects stably.

(a)

(b)