Query2Uncertainty uses density estimation on object query features to recalibrate both classification and regression uncertainties in 3D object detection, outperforming standard post-hoc methods in in-distribution and distribution-shifted scenarios.
Obtaining calibrated probability estimates from decision trees and naive Bayesian classifiers
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Query2Uncertainty: Robust Uncertainty Quantification and Calibration for 3D Object Detection under Distribution Shift
Query2Uncertainty uses density estimation on object query features to recalibrate both classification and regression uncertainties in 3D object detection, outperforming standard post-hoc methods in in-distribution and distribution-shifted scenarios.