Research map / Detection and segmentation / 3D detection and driving
3D Bounding Box Estimation Using Deep Learning and Geometry
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace single-bin/L2 orientation regression with 2-bin MultiBin regression (ablation) | Our Method · KITTI cars | OS: 0.89 -> 0.98 | Helped |
| Use 8-bin rather than 1-bin MultiBin regression (ablation) | Our Method · Pascal3D+ | Acc π/6: 0.65 -> 0.81 | Helped |
| Increase Pascal3D+ orientation bins from 8 to 16 (ablation) | Our Method · Pascal3D+ | Acc π/6: 0.81 -> 0.77 | Hurt |
| Increase KITTI orientation bins from 2 to 16 (ablation) | Our Method · KITTI cars | OS: 0.98 -> 0.96 | Hurt |
| Increase FC width from 64 to 1024 (ablation) | Our Method · KITTI cars | OS: 0.9583 -> 0.9861 | Helped |
| Use average category dimensions instead of regressing dimensions (ablation) | Our Method · KITTI cars | 3D bounding-box accuracy | Hurt |
| Reduce training instances from 10828 to 1100 | Our Method · KITTI cars | OS: 0.9808 -> 0.9026 | Hurt |
| Regress object dimensions and orientation, then solve translation using 2D-box geometry | Our Method · KITTI cars, Easy | AOS: 91.44% -> 92.90% | Helped |
| Use the proposed approach rather than SubCNN | Our Method · KITTI cars, Moderate | AOS: 88.62% -> 88.75% | Helped |
| Use the proposed approach rather than SubCNN | Our Method · KITTI cars, Hard | AOS: 78.68% -> 76.76% | Hurt |
| Use MultiBin viewpoint estimation rather than the listed prior method | Our Method · Pascal3D+, mean | MedErr: 13.6 -> 11.1 | Helped |
| Use MultiBin viewpoint estimation rather than the listed prior method | Our Method · Pascal3D+, mean | Acc π/6: 0.8075 -> 0.8103 | Helped |
Closest papers by meaning
- PerspectiveNet: 3D Object Detection from a Single RGB Image via Perspective Points (NeurIPS 2019 · similarity 0.90)
- HVPR: Hybrid Voxel-Point Representation for Single-Stage 3D Object Detection (CVPR 2021 · similarity 0.89)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.89)
- Triangulation Learning Network: From Monocular to Stereo 3D Object Detection (CVPR 2019 · similarity 0.89)
- Rethinking Geometric Depth in Monocular 3D Object Detection: A Projection-Consistent Reformulation (NeurIPS 2026 · similarity 0.89)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · 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.