Faults Visualization 

Fault Detection Visualization

In this subsection, we visualize several error-revealing test data generated by Drivence. 

We begin by presenting the overall tracking visualization results for the generated driving sequences, followed by a detailed analysis of the tracking errors through three case studies. Lastly, we provide the overall tracking results for several additional driving sequences.

Note that we provide generated image and point clouds with labeled bounding boxes. We leverge green bounding boxes to mark the ground-truth and red bounding boxes to mark the tracking results. 

Cases

In this subsection, we provide a detailed description of the tracking results

Case1

The tracking results of VirTrack on generated sequences (derived from the S1 scenario).

Driving Sequence

Errors

(1)  Partly Tracked: Car_2 (ID:2 in first frame), tracking_rate = 64.3%

(2) ID Switch: Car_0 (ID:0 in first frame) -> Car_28 (ID:28 in 41-th frame) 

(3) Fragmentation: Car_0 (tracked in 32th frame) -> Car_0 ( untracked in 33-th frame) 

Case2

The tracking results of YONTD on generated sequences (derived from the S1 scenario).

(1)  Partly Tracked: Car_3 (ID:3 in first frame), tracking_rate = 66.7%

        Partly Tracked: Car_9 (ID:9 in 5-th frame ), tracking_rate = 51.4%

(2) ID Switch: Car_3 (ID:3 in first frame) -> Car_14 (ID:14 in 21st frame) 

      ID Switch: Car_0 (ID:0 in 1-th frame) -> Car_24 (ID:24 in 37-th frame) 

      ID Switch: Car_9 (ID:9 in 10-th frame) -> Car_15 (ID:15 in 44-th frame) 

      ID Switch: Car_1 (ID:1 in first frame) -> Car_29 (ID:29 in 44-th frame) 

(3) Fragmentation: Car_3 (tracked in 5-th frame) -> Car_3( untracked in 6-th frame) 

     Fragmentation: Car_1 (tracked in 29-th frame) -> Car_1( untracked in 30-th frame) 

Case3

The tracking results of  VirTrack on generated sequences (derived from the S8 scenario).

(1)  Lose Tracked: 

Car_① , tracking_rate = 0  

Car_② , tracking_rate =0

(2)  Partly Tracked: Car_11 (ID:11 in first frame), tracking_rate = 67.7%

(3) ID Switch: The ID:5 is assigned to two cars.

Car_5 (labeled black car on the right of the road in second frame) -> Car_5 (labeled white car on the left of the road in 39-th frame) 

Overview  tracking results

We also provide an overview of several tracking results.

The tracking results of  VirTrack on generated sequences (derived from the S2 scenario).

The tracking results of YONTD on generated sequences (derived from the S2 scenario).

The tracking results of YONTD on generated sequences (derived from the S4 scenario).

The tracking results of  YONTD on generated sequences (derived from the S7 scenario).