Data from Ricco and Tomasi (CVPR 2012)

Download zip file here: riccoCVPR2012.zip

The zip file contains the input frames for the two non-synthetic sequence used in our CVPR 2012 paper: flowerbed and marple along with the trajectories returned by our algorithm and the public implementation of Sundaram et al. (ECCV 2010). We also include the sparse, hand-tracked, ground truth trajectories for the two sequences. Finally, we provide a Matlab function to compute the error metrics reported in our paper. To compute error metrics for your own results, add in a structure containing the X and Y coordinates of your estimated trajectories and an interpolated image stack formed by warping all images back to the first frame. See the included README.txt for more details.

For more information, please refer to our 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]