Research map / Detection and segmentation / 3D detection and driving
Kinematic 3D Object Detection in Monocular Video
Garrick Brazil, Gerard Pons-Moll, Xiaoming Liu, Bernt Schiele
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 orientation decomposition (axis, heading, offset) (ablation) | baseline detector · KITTI val, Car IoU 0.7 | AP3D Moderate: 9.71 -> 12.10 | Improved |
| add self-balancing confidence loss (ablation) | detector with orientation decomposition · KITTI val, Car IoU 0.7 | AP3D Moderate: 12.10 -> 12.77 | Improved |
| fuse self-balancing confidence with classification score (mu = c * omega) (ablation) | detector with confidence loss · KITTI val, Car IoU 0.7 | AP3D Moderate: 12.77 -> 13.55 | Improved |
| add kinematic Kalman motion model over 4 frames (ablation) | detector with decomposition and confidence · KITTI val, Car IoU 0.7 | AP3D Moderate: 13.55 -> 14.10 | Improved |
| use orientation decomposition instead of 10-bin classification (ablation) | orientation head · KITTI val, Car IoU 0.7 | AP3D Moderate: 10.74 -> 12.10 | Improved |
| use orientation decomposition instead of 2-bin classification (ablation) | orientation head · KITTI val, Car IoU 0.7 | AP3D Moderate: 9.47 -> 12.10 | Improved |
| use decomposed orientation instead of erratic baseline orientation inside the kinematic model (ablation) | kinematic model · KITTI val, Car IoU 0.7 | AP3D Moderate: 11.47 -> 14.10 | Improved |
| keep orientation decomposition and uncertainty in the kinematic model (vs baseline orientation and constant mu) (ablation) | kinematic model · KITTI val, Car IoU 0.7 | AP3D Moderate: 10.64 -> 14.10 | Improved |
| add orientation decomposition and kinematics (ablation) | baseline detector · KITTI val, Car IoU 0.7 | mean angle error (deg): 13.4 -> 10.9 -> 6.1 | Improved |
| use mu = c * omega instead of classification score c (ablation) | box rating · KITTI val, Car IoU 0.7 | correlation with 3D IoU: 0.301 -> 0.417 | Improved |
| forecast 1 frame ahead instead of current frame (ablation) | kinematic Kalman model · KITTI val, Car IoU 0.7 | AP3D Moderate: 14.10 -> 10.64 | Hurt |
| forecast 2 frames ahead instead of current frame (ablation) | kinematic Kalman model · KITTI val, Car IoU 0.7 | AP3D Moderate: 14.10 -> 5.10 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- M3D-RPN — M3D-RPN: Monocular 3D Region Proposal Network for Object Detection (ICCV 2019) · better on 6 of 6 shared results
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
- Time3D: End-to-End Joint Monocular 3D Object Detection and Tracking for Autonomous Driving (CVPR 2022 · similarity 0.91)
- Monocular 3D Object Detection with Depth from Motion (ECCV 2022 · similarity 0.90)
- PLOT: Pseudo-Labeling via Object Tracking for Monocular 3D Object Detection (ECCV 2026 · similarity 0.89)
- Joint Monocular 3D Vehicle Detection and Tracking (ICCV 2019 · similarity 0.88)
- Monocular 3D Object Detection: An Extrinsic Parameter Free Approach (CVPR 2021 · similarity 0.88)
- MonoDiff: Monocular 3D Object Detection and Pose Estimation with Diffusion Models (CVPR 2024 · similarity 0.87)
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.
