Research map / Detection and segmentation / 3D detection and driving
Objects Are Different: Flexible Monocular 3D Object Detection
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| decouple truncated-object localization and combine depth estimators using uncertainty | monocular 3D detection · KITTI test Car, moderate | AP3D|R40: 10.90 -> 13.89 | Helped |
| add separate inside/outside offset losses (ablation) | x_c/x_I object representations · KITTI val Car, moderate | AP3D|R40: 10.2 -> 11.7 | Helped |
| add edge fusion (ablation) | x_c/x_I object representations · KITTI val Car, hard | AP3D|R40: 8.9 -> 9.7 | Helped |
| use x_c for inside objects and x_I for outside objects, with decoupled loss and edge fusion (ablation) | monocular 3D detection · KITTI val Car, hard | AP3D|R40: 7.6 -> 11.4 | Helped |
| replace edge intersection x_I with clamped projected center x_cc (ablation) | outside-object representation · KITTI val Car, hard | AP3D|R40: 11.4 -> 8.6 | Hurt |
| replace edge intersection x_I with clamped 2D center x_cb (ablation) | outside-object representation · KITTI val Car, moderate | AP3D|R40: 12.6 -> 12.0 | Hurt |
| add uncertainty modeling to direct depth regression (ablation) | direct depth estimator · KITTI val Car, easy | AP3D|R40: 15.86 -> 19.63 | Helped |
| add uncertainty modeling to keypoint-based depth (ablation) | keypoint depth estimator · KITTI val Car, easy | AP3D|R40: 15.45 -> 18.42 | Helped |
| combine direct and keypoint-based depth estimates using uncertainty weights (ablation) | depth estimation · KITTI val Car, moderate | AP3D|R40: 14.83 -> 17.51 | Helped |
| replace soft uncertainty-weighted averaging with lowest-uncertainty hard selection (ablation) | depth ensemble · KITTI val Car, moderate | AP|R40: 17.51 -> 16.80 | Hurt |
| select the most accurate depth estimator per object using an oracle | depth ensemble · KITTI val Car, moderate | AP|R40: 17.51 -> 19.98 | Helped |
| use the proposed detector instead of MonoPair | KITTI 3D detection · KITTI test Pedestrian, moderate | AP3D|R40: 6.68 -> 6.31 | Hurt |
Closest papers by meaning
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.92)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.90)
- MonoMAE: Enhancing Monocular 3D Detection through Depth-Aware Masked Autoencoders (NeurIPS 2024 · similarity 0.89)
- FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 2024 · similarity 0.88)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.88)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · 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.