Research map / Detection and segmentation / 3D detection and driving
Depth-discriminative Metric Learning for Monocular 3D Object Detection
Wonhyeok Choi, Mingyu Shin, Sunghoon Im
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 quasi-isometric loss and object-wise depth loss | MonoCon · KITTI val Car Easy | AP_3D|R40: 23.03 -> 27.90 | Improved |
| add both losses | DID-M3D (LiDAR) · KITTI val Car Moderate | AP_3D|R40: 16.22 -> 17.12 | Improved |
| add quasi-isometric loss only (ablation) | MonoCon · KITTI val | overall AP gain: +8.7% | Improved |
| add object-wise depth loss only (ablation) | MonoCon · KITTI val | overall AP gain: +5.0% | Improved |
| add quasi-isometric loss only (ablation) | DID-M3D · KITTI val | overall AP gain: +2.3% | Improved |
| use SupCR instead of quasi-isometric loss (ablation) | MonoCon · KITTI val Car Moderate | AP_3D|R40: 17.84 -> 17.55 | Hurt |
| use SupCR contrastive regression (ablation) | MonoCon · KITTI val | E_dim: 0.025 -> 0.031 | Hurt |
| apply to anchor-based MonoDTR | MonoDTR · KITTI val Car Moderate | AP_3D|R40: 18.45 -> 19.07 | Improved |
| apply to BEV ImVoxelNet | ImVoxelNet · KITTI val Car Moderate | AP_3D|R40: 17.80 -> 18.20 | Improved |
| K = 2.0 instead of 1.5 (ablation) | quasi-isometric loss on MonoCon · KITTI val Car Moderate | AP_3D|R40: 18.68 -> 17.44 | Hurt |
| K = 1.0 instead of 1.5 (ablation) | quasi-isometric loss on MonoCon · KITTI val Car Moderate | AP_3D|R40: 18.68 -> 18.09 | Hurt |
| epsilon = infinity instead of 10 (ablation) | quasi-isometric loss on MonoCon · KITTI val Car Moderate | AP_3D|R40: 18.68 -> 16.91 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DID-M3D — DID-M3D: Decoupling Instance Depth for Monocular 3D Object Detection (ECCV 2022) · better on 3 of 3 shared results
- MonoDTR — MonoDTR: Monocular 3D Object Detection With Depth-Aware Transformer (CVPR 2022) · better on 1 of 1 shared result
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
- Geometry-Based Distance Decomposition for Monocular 3D Object Detection (ICCV 2021 · similarity 0.90)
- Objects Are Different: Flexible Monocular 3D Object Detection (CVPR 2021 · similarity 0.89)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.89)
- Disentangling Monocular 3D Object Detection (ICCV 2019 · similarity 0.89)
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.89)
- MonoMAE: Enhancing Monocular 3D Detection through Depth-Aware Masked Autoencoders (NeurIPS 2024 · similarity 0.89)
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.
