Figure 1.
The flowchart illustrates the steps involved in analysing video frames for object detection and tracking. It begins with video frames, which undergo object detection to produce bounding boxes. These bounding boxes are then used for object tracking, resulting in trajectories. The trajectories are processed using data-driven thresholding to generate stopped-on-tracks results, indicating when detected objects stop moving within the tracked video sequence.Stopped-on-tracks inference detection framework
Source(s): Created by authors