Research map / Detection and segmentation / 3D detection and driving
Monocular 3D Object Detection via Feature Domain Adaptation
Xiaoqing Ye, Liang Du, Yifeng Shi, Yingying Li, Xiao Tan, Jianfeng Feng, Errui Ding, Shilei Wen
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 feature domain adaptation (pseudo-LiDAR to LiDAR features) (ablation) | baseline · KITTI val Car Moderate | AP3D: 21.9 -> 23.1 | Improved |
| add supervised foreground segmentation loss (ablation) | baseline without DA · KITTI val Car Moderate | AP3D: 15.7 -> 21.9 | Improved |
| add context-aware foreground segmentation (CAFS) (ablation) | baseline without DA · KITTI val Car Moderate | AP3D: 21.9 -> 22.4 | Improved |
| replace supervised segmentation loss with CAFS (ablation) | DA model · KITTI val Car Moderate | AP3D: 23.1 -> 23.6 | Improved |
| add foreground segmentation supervision/CAFS over unsupervised segmentation (ablation) | DA model · KITTI val Car Moderate | AP3D: 16.1 -> 23.6 | Improved |
| use two-branch decoder (xyzr, whl) instead of one (ablation) | DA + CAFS · KITTI val Car Moderate | AP3D: 23.6 -> 24.0 | Improved |
| use three-branch decoder instead of two (ablation) | DA + CAFS · KITTI val Car Moderate | AP3D: 24.0 -> 23.2 | Hurt |
| two-branch decoder without DA (ablation) | baseline · KITTI val Car Moderate | AP3D: 21.9 -> 22.1 | Improved |
| use L2 instead of L1 adaptation loss (ablation) | DA-3Ddet · KITTI val Car Moderate | AP3D: 23.3 -> 24.0 | Improved |
| increase sampled points from 1024 (ablation) | DA-3Ddet · KITTI val Car Moderate | AP3D: 24.0 -> 23.8 | No clear effect |
| adapt mono features to LiDAR features | Mono pseudo-LiDAR · KITTI val Car Moderate | AP3D: 21.9 -> 24.0 | Improved |
| adapt stereo features to LiDAR features | Stereo pseudo-LiDAR · KITTI val Car Moderate | AP3D: 44.0 -> 46.1 | 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 2 of 2 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
- MonoDistill: Learning Spatial Features for Monocular 3D Object Detection (ICLR 2022 · similarity 0.92)
- FD3D: Exploiting Foreground Depth Map for Feature-Supervised Monocular 3D Object Detection (AAAI 2024 · similarity 0.90)
- Geometry-Guided Domain Generalization for Monocular 3D Object Detection (AAAI 2024 · similarity 0.89)
- MonoTAKD: Teaching Assistant Knowledge Distillation for Monocular 3D Object Detection (CVPR 2025 · similarity 0.88)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.88)
- CMDA: Cross-Modal and Domain Adversarial Adaptation for LiDAR-Based 3D Object Detection (AAAI 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.
