Research map / Detection and segmentation / 3D detection and driving
MonoDistill: Learning Spatial Features for Monocular 3D Object Detection
Zhiyu Chong, Xinzhu Ma, Hong Zhang, Yuxin Yue, Haojie Li, Zhihui Wang, Wanli Ouyang
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 |
|---|---|---|---|
| Distill spatial cues from an aligned LiDAR teacher into MonoDLE | MonoDLE · KITTI test, Car 3D@IoU=0.7 | AP|R40: 12.26 -> 16.03 | Improved |
| Add scene-level feature distillation (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 15.13 -> 16.96 | Improved |
| Add object-level feature distillation (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 15.13 -> 16.85 | Improved |
| Add object-level result distillation (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 15.13 -> 17.24 | Improved |
| Combine scene-level, object-level feature, and object-level result distillation (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 15.13 -> 18.24 | Improved |
| Add attention-based feature fusion to the three distillation schemes (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 18.24 -> 18.47 | Improved |
| Restrict object-level feature distillation to foreground regions rather than the whole image (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 16.13 -> 16.85 | Improved |
| Use diffused region labels rather than sparse point labels for result distillation (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 15.51 -> 17.24 | Improved |
| Use LiDAR-guided knowledge distillation instead of direct dense-depth supervision | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 17.05 -> 18.47 | Improved |
| Use LiDAR rather than stereo depth to guide the teacher (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 18.18 -> 18.47 | Improved |
| Use dense rather than sparse LiDAR maps for teacher training (ablation) | MonoDLE · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 18.07; 18.47 | No clear effect |
| Apply MonoDistill to GUPNet | GUPNet · KITTI validation, Car 3D@IoU=0.7 moderate | AP|R40: 16.46 -> 16.69 | Improved |
Later papers that compared with this method
1 later accepted paper reports MonoDistill as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- BEVDistill: Cross-Modal BEV Distillation for Multi-View 3D Object Detection (ICLR 2023 · 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
- MonoTAKD: Teaching Assistant Knowledge Distillation for Monocular 3D Object Detection (CVPR 2025 · similarity 0.91)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.90)
- Training an Open-Vocabulary Monocular 3D Detection Model without 3D Data (NeurIPS 2024 · similarity 0.90)
- FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 2024 · similarity 0.90)
- MonoCD: Monocular 3D Object Detection with Complementary Depths (CVPR 2024 · similarity 0.90)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.90)
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.
