Abstract

We present a continuous optimization framework for inter- active tracking of 2D generic ob jects in a single video stream. The user begins with specifying the locations of a target ob ject in a small set of keyframes; the system then automatically tracks locations of the ob jects by combining user constraints with visual measurements across the en- tire sequence. We formulate the problem in a spacetime optimization framework that optimizes over the whole sequence simultaneously. The resulting solution is consistent with visual measurements across the en- tire sequence while satisfying user constraints. We also introduce prior terms to reduce tracking ambiguity. We demonstrate the power of our algorithm on tracking ob jects with significant occlusions, scale and orien- tation changes, illumination changes, sudden movement of ob jects, and also simultaneous tracking of multiple ob jects. We compare the perfor- mance of our algorithm with alternative methods.