Problem
Research questions and friction points this paper is trying to address.
3D semantic occupancy prediction
out-of-distribution detection
long-tailed class bias
autonomous driving
ProOOD: Prototype-Guided Out-of-Distribution 3D Occupancy Prediction
🤖 AI Summary
This work addresses the challenges of anomalous misclassifications and overconfidence in 3D semantic occupancy prediction caused by long-tailed class distributions and out-of-distribution (OOD) inputs. The authors propose a lightweight, plug-and-play approach that, for the first time, integrates prototype-guided mechanisms into 3D occupancy prediction. By combining semantic inpainting, tail-class mining, and an EchoOOD scoring strategy that requires no additional training, the method achieves robust voxel-level OOD detection and calibration through the fusion of local logit consistency and local-to-global prototype matching. Evaluated on SemanticKITTI, the approach improves overall mIoU by 3.57% and boosts tail-class mIoU by 24.80%; on VAA-KITTI, it increases AuPRCr by 19.34 points, outperforming existing methods across multiple metrics.
📝 Abstract
3D semantic occupancy prediction is central to autonomous driving, yet current methods are vulnerable to long-tailed class bias and out-of-distribution (OOD) inputs, often overconfidently assigning anomalies to rare classes. We present ProOOD, a lightweight, plug-and-play method that couples prototype-guided refinement with training-free OOD scoring. ProOOD comprises (i) prototype-guided semantic imputation that fills occluded regions with class-consistent features, (ii) prototype-guided tail mining that strengthens rare-class representations to curb OOD absorption, and (iii) EchoOOD, which fuses local logit coherence with local and global prototype matching to produce reliable voxel-level OOD scores. Extensive experiments on five datasets demonstrate that ProOOD achieves state-of-the-art performance on both in-distribution 3D occupancy prediction and OOD detection. On SemanticKITTI, it surpasses baselines by +3.57% mIoU overall and +24.80% tail-class mIoU; on VAA-KITTI, it improves AuPRCr by +19.34 points, with consistent gains across benchmarks. These improvements yield more calibrated occupancy estimates and more reliable OOD detection in safety-critical urban driving. The source code is publicly available at https://github.com/7uHeng/ProOOD.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Prototype-Guided
Out-of-Distribution Detection
3D Occupancy Prediction
Tail-Class Enhancement
Training-Free OOD Scoring
🔎 Similar Papers
No similar papers found.
