Research map / Detection and segmentation / 3D detection and driving
LeAD-M3D: Leveraging Asymmetric Distillation for Real-Time Monocular 3D Detection
Johannes Meier, Jonathan Michel, Oussema Dhaouadi, Yung-Hsu Yang, Christoph Reich, Zuria Bauer, Stefan Roth, Marc Pollefeys, Jacques Kaiser, Daniel Cremers
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 3D MGIoU term to matching (CM3D) (ablation) | YOLOv10-M3D B · KITTI val Car | AP3D Mod.: 19.60 -> 20.43 | Improved |
| add A2D2 distillation (ablation) | YOLOv10-M3D B with CM3D · KITTI val Car | AP3D Mod.: 20.43 -> 22.65 | Improved |
| distill depth features instead of backbone features (ablation) | LeAD-M3D B · KITTI val Car | AP3D Mod.: 22.30 -> 22.65 | Improved |
| add teacher-quality weight (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 22.28 -> 22.65 | Improved |
| add channel-importance weight (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 22.04 -> 22.65 | Improved |
| use offline instead of online distillation (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 21.80 -> 22.65 | Improved |
| feed teacher clean images (asymmetry) rather than the same MixUp image (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 21.57 -> 22.65 | Improved |
| use MixUp instead of CutMix or RandAugment for student input (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 21.10/22.28 -> 22.65 | Improved |
| any augmentation-based distillation vs no distillation (ablation) | A2D2 · KITTI val Car | AP3D Mod.: 20.43 -> 21.10 (RandAugment) | Improved |
| add 3D matching when A2D2 is used (ablation) | CM3D · KITTI val Car | AP3D Easy/Mod.: 27.12 -> 28.44 Easy; 21.81 -> 22.65 Mod. | Improved |
| add confidence-gated 3D inference (ablation) | LeAD-M3D N · KITTI val Car | GFLOPs / runtime: 81.0 -> 19.5 GFLOPs; 16.2 -> 10.9 ms; AP 18.33 unchanged | Improved |
| reduce head channels from 128 to 64 (ablation) | LeAD-M3D N · KITTI val Car | AP3D Mod.: 18.33 -> 18.38 | No clear effect |
Closest papers by meaning
- Weakly Supervised Monocular 3D Detection with a Single-View Image (CVPR 2024 · similarity 0.90)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.90)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.90)
- 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)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · 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.
