Research map / Detection and segmentation / 3D detection and driving
Difficulty-Aware Label-Guided Denoising for Monocular 3D Object Detection
Soyul Lee, Seungmin Baek, Dongbo Min
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 difficulty-aware label-guided denoising (MonoDLGD) | MonoDGP · KITTI val Car AP3D Moderate | AP3D R40: 22.34 -> 25.19 | Improved |
| add the method on top of MonoDETR | MonoDETR · KITTI val Car AP3D Easy | AP3D R40: 27.34 -> 29.79 | Improved |
| swap anchor queries for 3D-DAB without denoising supervision (ablation) | MonoDGP · KITTI val Car AP3D Moderate | AP3D R40: 22.34 -> 20.64 | Hurt |
| add uniform label perturbation to 3D-DAB (ablation) | 3D-DAB · KITTI val Car AP3D Moderate | AP3D R40: 20.64 -> 23.82 | Improved |
| add Laplacian uncertainty loss to uniform noise denoising (ablation) | uniform noise denoising · KITTI val Car AP3D Moderate | AP3D R40: 23.82 -> 24.7 | Improved |
| replace uniform noise with difficulty-aware perturbation (ablation) | 3D-DAB + LU · KITTI val Car AP3D Moderate | AP3D R40: 24.7 -> 25.19 | Improved |
| add depth to denoising targets (ablation) | Bbox + Class denoising · KITTI val Car AP3D Moderate | AP3D R40: 23.36 -> 25.19 | Improved |
| use ground-truth labels instead of uniform-noised labels in stage 1 (ablation) | DAP stage 1 · KITTI val Car AP3D Moderate | AP3D R40: 23.85 -> 25.19 | Improved |
| add EMA to difficulty min/max normalization (beta 0.8) (ablation) | DAP · KITTI val Car AP3D Moderate | AP3D R40: 24.32 -> 25.19 | Improved |
| vary EMA momentum to 0.9 (ablation) | DAP · KITTI val Car AP3D Moderate | AP3D R40: 25.19 -> 24.29 | Hurt |
| increase perturbation groups from 7 to 8 (ablation) | MonoDLGD · KITTI val Car AP3D Moderate | AP3D R40: 25.19 -> 23.93 | Hurt |
| vary bbox scale factor gamma^b from 0.4 to 0.2 (ablation) | DAP · KITTI val Car AP3D Easy | AP3D R40: 34.89 -> 32.02 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- MonoDGP — MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025) · better on 7 of 7 shared results
- MonoDETR — MonoDETR: Depth-guided Transformer for Monocular 3D Object Detection (ICCV 2023) · better on 6 of 6 shared results
- FD3D — FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 2024) · better on 3 of 3 shared results
- MonoMAE — MonoMAE: Enhancing Monocular 3D Detection through Depth-Aware Masked Autoencoders (NeurIPS 2024) · better on 3 of 3 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
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.89)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.88)
- MonoMAE: Enhancing Monocular 3D Detection through Depth-Aware Masked Autoencoders (NeurIPS 2024 · similarity 0.88)
- Geometry-Guided Domain Generalization for Monocular 3D Object Detection (AAAI 2024 · similarity 0.88)
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.87)
- FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 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.
