Research Article
SKT-MOT and DyTracker: A Multiobject Tracking Dataset and a Dynamic Tracker for Speed Skating Video
| Input: Video length ; | | Number of tracks ; | | Initial frame tracks number; | | instability; | | 1 | for frame todo | | 2 | ifis instability then | | 3 | ifpresence ofdetections not matched to any track and their confidence are enough high then | | 4 | Generate new tracks; | | 5 | ; | | 6 | end | | 7 | else ifpresence oftracks not matched to any detection for fifteen consecutive frames then | | 8 | Delete these tracks; | | 9 | ; | | 10 | else ifno change for fifteen consecutive frames then | | 11 | stability; | | 12 | end | | 13 | else ifis stability then | | 14 | ifpresence ofdetections not matched to any track for fifteen consecutive frames then | | 15 | Generate new tracks; | | 16 | ; | | 17 | end | | 18 | else ifpresence oftracks not matched to any detection for fifteen consecutive frames then | | 19 | Delete these tracks; | | 20 | ; | | 21 | else | | 22 | remain unchanged; | | 23 | end |
|
Algorithm 1: TDM. |