We provide the following dataset for comparison purposes. If you use this data, please cite our paper.
We also provide code to load/manipulate the data on github.
After publication of the TVCG paper, we have tried replacing the convolution layers of the network with the SqueezeNet architecture. Doing so results in much greater robustness to occlusions at the expense of being slightly less stable, as demonstrated by the following occlusions graphs:
The authors gratefully acknowledge the following funding sources: