Research map / Detection and segmentation / 3D detection and driving
The Devil Is in the Task: Exploiting Reciprocal Appearance-Localization Features for Monocular 3D Object Detection
Zhikang Zou, Xiaoqing Ye, Liang Du, Xianhui Cheng, Xiao Tan, Li Zhang, Jianfeng Feng, Xiangyang Xue, Errui Ding
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 self-reflect (S-R) module (ablation) | M3D-RPN baseline · KITTI val1 Car (R40) | AP_3D Mod.: 11.07 -> 13.08 | Improved |
| add self-reflect module on top of mutual-reflect (ablation) | M3D-RPN + M-R · KITTI val1 Car (R40) | AP_3D Mod.: 13.01 -> 13.39 | Improved |
| combine S-R and M-R rather than either alone (ablation) | ALFR · KITTI val1 Car (R40) | AP_3D Mod.: 13.08 -> 13.39 | Improved |
| add DIT module to ALFR (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.39 -> 14.79 | Improved |
| assign rotation and dimensions to appearance stream instead of localization (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.35 -> 14.79 | Improved |
| assign rotation to appearance rather than localization task (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.35 -> 14.31 | Improved |
| add appearance-to-localization information flow (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.46 -> 13.64 | Improved |
| add localization-to-appearance information flow (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.46 -> 13.73 | Improved |
| use fixed initialized trading scores (DIT-init) instead of no DIT (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.39 -> 14.07 | Improved |
| generate trading scores cross-task (DIT-cross) (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.39 -> 12.44 | Hurt |
| add DIT scores from the task's own features instead of shared features (ablation) | DFR-Net · KITTI val1 Car (R40) | AP_3D Mod.: 13.87 -> 14.79 | Improved |
| apply DFR-Net to M3D-RPN backbone | M3D-RPN · KITTI val1 Car | AP_3D Easy: 14.53 -> 19.55 | Improved |
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 5 of 5 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
- MonoDETR: Depth-guided Transformer for Monocular 3D Object Detection (ICCV 2023 · similarity 0.88)
- A Depth Complementary Framework for Monocular 3D Object Detection in Autonomous Driving (TPAMI 2026 · similarity 0.88)
- Triangulation Learning Network: From Monocular to Stereo 3D Object Detection (CVPR 2019 · similarity 0.88)
- Kinematic 3D Object Detection in Monocular Video (ECCV 2020 · similarity 0.88)
- MonoTAKD: Teaching Assistant Knowledge Distillation for Monocular 3D Object Detection (CVPR 2025 · similarity 0.87)
- SparseAlign: a Fully Sparse Framework for Cooperative Object Detection (CVPR 2025 · similarity 0.86)
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.
