Dense Lagrangian Motion Estimation with Occlusions
(CVPR 2012)

NEW: Our data are now available.

We couple occlusion modeling and multi-frame motion estimation to compute dense, temporally extended point trajectories in video with significant occlusions. Our approach combines robust spatial regularization with spatially and temporally global occlusion labeling in a variational, Lagrangian framework with subspace constraints. Our approach allows us to maintain accurate tracks across temporary occlusions -- we return a trajectory for each world point that lists its position in every frame in which it is visible.

For more, see our poster from CVPR or read the paper.

S. Ricco and C. Tomasi. Dense Lagrangian Motion Estimation with Occlusions. 25th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2012). Providence, Rhode Island, USA, June 2012. [PDF]