Anonymous submission — under double-blind review. Author and affiliation details are withheld.
← RareOcc project page  ·  λ=0.05 (primary)

Occupancy generation vs. UniScene-v2 — λ=0.15

These are the results at the paper's setting, λ=0.15. After submission we found the released UniScene-v2 code uses λ=0.05 by default; those (marginally different) results are on the λ=0.05 page and are the ones we now consider primary. ROCC's margin is unchanged in substance at either setting.

UniScene-v2UniScene-v2Dense+ bgROCC
SetMetric(λ=0.15)(λ=0)CNNhead‡(ours)
mini-valfg-mIoU ↑18.0733.5545.6249.25
mIoUUS25.8138.0949.0151.65
occ-IoU ↑16.6424.7424.1927.25
key subsetfg-mIoU ↑18.050.0332.9144.9048.45
mIoUUS25.8713.1237.5148.3550.92
occ-IoU ↑16.567.3123.9023.4026.17

Occupancy generation on two views of the nuPlan-Occ validation split: mini-val (the full validation set) and key subset (one key frame per clip). One shared per-class IoU accumulator scores every column on identical ground truth. UniScene-v2's λ supplies a fraction of the ground-truth occupancy latent of the scored frame; at λ=0 it collapses. ‡ trained at a larger effective batch than the other columns.