Abstract
A method for online, real-time tracking of ob jects is pre- sented. Tracking is treated as a repeated detection problem where po- tential target ob jects are identified with a pre-trained category detector and ob ject identity across frames is established by individual-specific de- tectors. The individual detectors are (re-)trained online from a single positive example whenever there is a coincident category detection. This ensures that the tracker is robust to drift. Real-time operation is possi- ble since an individual-ob ject detector is obtained through elementary manipulations of the thresholds of the category detector and therefore only minimal additional computations are required. Our tracking algo- rithm is benchmarked against nine state-of-the-art trackers on two large, publicly available and challenging video datasets. We find that our al- gorithm is 10% more accurate and nearly as fast as the fastest of the competing algorithms, and it is as accurate but 20 times faster than the most accurate of the competing algorithms.