Research map / Point clouds / Segmentation and detection
Accurate Monocular 3D Object Detection via Color-Embedded 3D Reconstruction for Autonomous Driving
Xinzhu Ma, Zhihui Wang, Haojie Li, Pengbo Zhang, Wanli Ouyang, Xin Fan
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 RGB fusion module (vs no RGB) (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 19.46 -> 21.09 | Improved |
| add RGB fusion module (vs no RGB) (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D localization, moderate | APloc (%): 26.28 -> 28.39 | Improved |
| use attention fusion instead of plain concatenation of RGB (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 18.25 -> 21.09 | Improved |
| plain concatenation of RGB vs using no RGB (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 19.46 -> 18.25 | Hurt |
| add prior-based points segmentation (vs none) (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate, IoU=0.7 | AP3D (%): 18.22 -> 21.09 | Improved |
| add prior-based points segmentation (vs none) (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate, IoU=0.5 | AP3D (%): 44.53 -> 49.19 | Improved |
| prior-based segmentation instead of segmentation PointNet (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate, IoU=0.7 | AP3D (%): 18.70 -> 21.09 | Improved |
| vary segmentation threshold r from -0.5 to 1.0 m (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 20.01, 20.55, 21.09, 20.93 | No clear effect |
| use more accurate depth generator (MonoDepth to PSMNet) (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 15.45 -> 26.03 | Improved |
| DORN instead of MonoDepth depth generator (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 15.45 -> 21.09 | Improved |
| increase sampled points from 64 to 512 (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 19.41 -> 21.09 | Improved |
| increase sampled points from 512 to 1024 (ablation) | Color-embedded 3D reconstruction detector · KITTI val Car 3D detection, moderate | AP3D (%): 21.09 -> 21.01 | No clear effect |
Closest papers by meaning
- Multi-View 3D Object Detection Network for Autonomous Driving (CVPR 2017 · similarity 0.91)
- GS3D: An Efficient 3D Object Detection Framework for Autonomous Driving (CVPR 2019 · similarity 0.91)
- M3D-RPN: Monocular 3D Region Proposal Network for Object Detection (ICCV 2019 · similarity 0.90)
- RTM3D: Real-time Monocular 3D Detection from Object Keypoints for Autonomous Driving (ECCV 2020 · similarity 0.90)
- Monocular 3D Object Detection for Autonomous Driving (CVPR 2016 · similarity 0.90)
- Training an Open-Vocabulary Monocular 3D Detection Model without 3D Data (NeurIPS 2024 · 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.
