Research map / Detection and segmentation / 3D detection and driving
Geometry Uncertainty Projection Network for Monocular 3D Object Detection
Yan Lu, Xinzhu Ma, Lei Yang, Tianzhu Zhang, Yating Liu, Qi Chu, Junjie Yan, Wanli Ouyang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add coordinate map to RoI features (ablation) | GUP Net baseline · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 11.00 -> 12.44 | Improved |
| add uncertainty-confidence (UnC) to CM model (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): +1.09 | Improved |
| add uncertainty-confidence (UnC) to CM+GeP model (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): +0.78 | Improved |
| add geometry projection (GeP) without UnC (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 12.44 -> 12.79 | Improved |
| add geometry projection (GeP) with UnC but without GeU (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 13.53 -> 13.57 | Mixed |
| add geometry-based uncertainty (GeU) (ablation) | GUP Net without HTL · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 13.57 -> 15.70 | Improved |
| add geometry-based uncertainty (GeU) (ablation) | GUP Net with HTL · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 15.63 -> 16.46 | Improved |
| add Hierarchical Task Learning (HTL) (ablation) | GUP Net without GeU · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 13.57 -> 15.63 | Improved |
| add Hierarchical Task Learning (HTL) (ablation) | GUP Net with GeU · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 15.70 -> 16.46 | Improved |
| replace GradNorm loss weighting with HTL (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 10.49 -> 16.46 | Improved |
| replace Task Uncertainty loss weighting with HTL (ablation) | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 13.94 -> 16.46 | Improved |
| use GUP Net instead of MonoPair at 0.7 IoU | GUP Net · KITTI val Car 3D IoU=0.7 Mod. | AP40 (%): 12.30 -> 16.46 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MonoPair — MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships (CVPR 2020) · better on 6 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GUPNet++: Geometry Uncertainty Propagation Network for Monocular 3D Object Detection (TPAMI 2024 · similarity 0.94)
- Rethinking Geometric Depth in Monocular 3D Object Detection: A Projection-Consistent Reformulation (NeurIPS 2026 · similarity 0.89)
- Triangulation Learning Network: From Monocular to Stereo 3D Object Detection (CVPR 2019 · similarity 0.88)
- MonoJSG: Joint Semantic and Geometric Cost Volume for Monocular 3D Object Detection (CVPR 2022 · similarity 0.88)
- Adaptive Dual Uncertainty Optimization: Boosting Monocular 3D Object Detection under Test-Time Shifts (ICCV 2025 · similarity 0.88)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
