Raw results?
Closed this issue · 3 comments
laisimiao commented
Could you provide the raw results like in your paper?
and do you evaluate the MonTrack on COESOT or CEUTrack on FE108?
ZHU-Zhiyu commented
@laisimiao Raw results are in here. Actually, we have tried many times to cross-adapt those trackers; however, their performance is unsatisfactory. It may be due to the huge distribution gaps between those datasets. If any further question, feel free to ask.
laisimiao commented
Thanks. And is the event input form of MonTrack voxel
? If not, it is trouble to re-convert COESOT dataset to its specific form and evaluate.
ZHU-Zhiyu commented
@laisimiao It takes raw event cloud as input, correponding processing code as here. Here is one of our results for adapting MonTrack on COESOT, code as here.