AI Paper Atlas

Research map / Benchmarks / Occ3D-nuScenes

Occ3D-nuScenes leaderboard

28 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Occ3D-nuScenes. Best mIoU (val): 55.1 (Gau-Occ, CVPR 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

Occ3D-nuScenes · mIoU (val)

16 papers, each with its best reported number for its own method; higher is better.

#mIoU ↑MethodPaperTopic
155.1Gau-OccGau-Occ: Geometry-Completed Gaussians for Multi-Modal 3D Occupancy Prediction
CVPR 2026 · Table 2
3D generation
253.01SurroundSDF Large
ConvNext-B
SurroundSDF: Implicit 3D Scene Understanding Based on Signed Distance Field
CVPR 2024 · Table 3
3D generation
351.66SDG-Fusion
ResNet-50
SDGOCC: Semantic and Depth-Guided Bird's-Eye View Transformation for 3D Multimodal Occupancy Prediction
CVPR 2025 · Table 1
3D generation
446.51REO (Ours)Robust 3D Semantic Occupancy Prediction With Calibration-Free Spatial Transformation
TPAMI 2026 · Table 5
3D generation
544.93CSV-Occ-B (4F)
R101
CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism
ICML 2025 · Table 1
3D detection and driving
644.1COTR (+Dr.Occ)
ResNet-50
Dr.Occ: Depth- and Region-Guided 3D Occupancy from Surround-View Cameras for Autonomous Driving
CVPR 2026 · Table 1
3D generation
744.02OctreeOcc
ResNet101
OctreeOcc: Efficient and Multi-Granularity Occupancy Prediction Using Octree Queries
NeurIPS 2024 · Table: 3D Occupancy prediction
3D generation
841.85ViewFormer
ResNet50
ViewFormer: Exploring Spatiotemporal Modeling for Multi-View 3D Occupancy Perception via View-Guided Transformers
ECCV 2024 · Table 1
3D detection and driving
941.7GSD-Occ*
ResNet-50
Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement
AAAI 2025 · Table 1
3D generation
1038.18ODG-L (8f)
ResNet-50
ODG: Occupancy Prediction Using Dual Gaussians
NeurIPS 2025 · Table 1
Gaussian splatting
6 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Occ3D-nuScenes · mIoU

15 papers, each with its best reported number for its own method; higher is better.

#mIoU ↑MethodPaperTopic
154.21RIOcc (R50, 256x704, C&L)
ResNet50
RIOcc: Efficient Cross-Modal Fusion Transformer with Collaborative Feature Refinement for 3D Semantic Occupancy Prediction
ICCV 2025 · Table 1
3D detection and driving
251.66Baseline + SDG + FOADSDGOCC: Semantic and Depth-Guided Bird's-Eye View Transformation for 3D Multimodal Occupancy Prediction
CVPR 2025 · Table 3
3D generation
346.2BEVDet4D + COTR
BEVDet4D, SwinTransformer-B
COTR: Compact Occupancy TRansformer for Vision-based 3D Occupancy Prediction
CVPR 2024 · Table 1
3D generation
444.64SA-Occ (V5)*
Swin-B, 2 frames
SA-Occ: Satellite-Assisted 3D Occupancy Prediction in Real World
ICCV 2025 · Table 7
3D generation
542.37Sandwich Eikonal with joint supervision (Model E)
SurroundSDF
SurroundSDF: Implicit 3D Scene Understanding Based on Signed Distance Field
CVPR 2024 · Table 5
3D generation
641.21BEVFormer w/ OSP (frozen backbone)
BEVFormer
Occupancy as Set of Points
ECCV 2024 · Table 1
3D generation
736.2OPUS-L (8f)
R50
OPUS: Occupancy Prediction Using a Sparse Set
NeurIPS 2024 · Table 1
RNNs and normalization
834.69PreWorld + pre-training
BEVStereo
Semi-Supervised Vision-Centric 3D Occupancy World Model for Autonomous Driving
ICLR 2025 · 3D occupancy prediction table
3D generation
932.82ODG-T with motion compensation, query attention, and rendering supervision
ResNet-50
ODG: Occupancy Prediction Using Dual Gaussians
NeurIPS 2025 · Table 4
Gaussian splatting
1028.3Vampire
ResNet-50
Regulating Intermediate 3D Features for Vision-Centric Autonomous Driving
AAAI 2024 · Table 1
3D generation
5 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.