Visualization 1 Final.mp4

Visualization 1 shows a sample video of the experimental results comparing 2D tracking and 3D tracking using the proposed approach. In this video, a green-colored box represents a valid detection using YOLOv2 deep neural network (detection score more than 0.5) and a red-colored box represents the 2D bounding box of the object corresponding to a failed detection. As we can see, 2D imaging fails sporadically in tracking the object in degraded environments. In comparison, 3D integral imaging-based tracking performs much better in similar circumstances.