Very nice work and thanks for releasing the dataset. So far I have one question regarding the source of the dataset.
Which scenes of the Replica correspond to the 101 training sequences (or 11 test sequences)?
For example, train's 00 should correspond to room_2, how about other scenes?