For my project, I need to non-rigidly register multiple scans. How can I use NDP for more than 2 scans?
You provide the _4dmatch_multiview.py dataset - is there also a config that can be used to train and test on this dataset?
Is the pairwise_overlap a binary flow mask?
What are the entries _2axis_flow, axis_node, and poses? Are they required?