Research map / Detection and segmentation / 3D detection and driving
Adaptive Dual Uncertainty Optimization: Boosting Monocular 3D Object Detection under Test-Time Shifts
Zixuan Hu, Dongxiao Li, Xinzhu Ma, Shixiang Tang, Xiaotong Li, Wenhan Yang, Ling-Yu Duan
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 conjugate focal loss to the source model (ablation) | MonoFlex · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 4.54 -> 20.98 | Improved |
| add normal consistency loss alone to the source model (ablation) | MonoFlex · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 4.54 -> 12.38 | Improved |
| add normal consistency loss to conjugate focal loss (ablation) | DUO (CFL) · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 20.98 -> 22.17 | Improved |
| add semantic guidance mask to CFL+NCL (ablation) | DUO (CFL+NCL) · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 22.17 -> 22.97 | Improved |
| add semantic guidance mask to NCL alone (ablation) | NCL · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 12.38 -> 16.49 | Improved |
| add conjugate focal loss to NCL+mask (ablation) | NCL + mask · KITTI-C Car (MonoFlex, Gaussian shift) | AP3D R40 Car: 16.49 -> 22.97 | Improved |
| add conjugate focal loss alone (MonoGround) (ablation) | MonoGround · KITTI-C Car (MonoGround, Gaussian shift) | AP3D R40 Car: 3.94 -> 22.89 | Improved |
| full DUO versus partial CFL+NCL (MonoGround) (ablation) | MonoGround · KITTI-C Car (MonoGround, Gaussian shift) | AP3D R40 Car: 24.29 -> 24.73 | Improved |
| replace MonoTTA with DUO | MonoFlex · KITTI-C level 5 Car | AP3D R40 avg: 20.87 -> 22.97 | Improved |
| replace MonoTTA with DUO | MonoGround · KITTI-C level 5 Car | AP3D R40 avg: 22.57 -> 24.73 | Improved |
| replace MonoTTA with DUO under mild corruption | MonoFlex · KITTI-C level 1 Car | AP3D R40 avg: 39.33 -> 39.66 | Improved |
| replace MonoTTA with DUO on real-world shift | MonoFlex · nuScenes Sunny<->Rainy | AP3D R40 avg: 11.01 -> 12.38 | 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.
- MonoTTA — MonoTTA: Fully Test-Time Adaptation for Monocular 3D Object Detection (ECCV 2024) · better on 10 of 10 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
- Cocoon: Robust Multi-Modal Perception with Uncertainty-Aware Sensor Fusion (ICLR 2025 · similarity 0.89)
- MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships (CVPR 2020 · similarity 0.89)
- Geometry-Guided Domain Generalization for Monocular 3D Object Detection (AAAI 2024 · similarity 0.88)
- Weakly Supervised Monocular 3D Object Detection by Spatial-Temporal View Consistency (TPAMI 2024 · similarity 0.88)
- Geometry Uncertainty Projection Network for Monocular 3D Object Detection (ICCV 2021 · similarity 0.88)
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.87)
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.
