Metrics

[EricHYD]

Hi @tovacinni, thx for your excellent work!
After i read your paper,you sample 131072 points for Chamfer distance (in B.2. Reconstruction Metrics).I want to figure out how many points sampled for gIou.
Thanks again.

[joeylitalien]

Hi @EricHYD,

I believe we used the same number of random points for gIoU. Anything above 100K should be sufficient to get a good estimate of the metric.

Best,
Joey