This dataset is similar to the [3D semantic segmentation dataset][1] but more targeted toward clustering technique development. In this dataset, partitions (i.e. clusters) of pixels are provided for every particle produced in simulation. This dataset contains individual electromagnetic (EM) particle produced in a cascade of EM shower while they are treated as one particle instance in the segmentation dataset. Further, directed inter-particle correlations (i.e. particle "flow") are recorded in particle information. This may be useful to develop particle clustering methods.
[1]: http://osf.io/vruzp