Research map / Detection and segmentation / 3D detection and driving
From Dataset to Real-world: General 3D Object Detection via Generalized Cross-domain Few-shot Learning
Shuangzhi Li, Junlong Shen, Lei Ma, Xingyu Li
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 image-guided multi-modal fusion (GDino+SAM) to prototype-learning finetuning (ablation) | GCFS (VoxelRCNN) · NuScenes->FS-KITTI 5-shot | overall mAP: 10.97 -> 13.55 | Improved |
| add contrastive-enhanced prototype learning to image fusion (ablation) | GCFS (VoxelRCNN) · NuScenes->FS-KITTI 5-shot | overall mAP: 12.68 -> 13.55 | Improved |
| add contrastive loss to learnable prototypes (ablation) | GCFS · NuScenes->FS-KITTI 5-shot, novel | mAP: 10.33 -> 11.72 | Improved |
| add LFVD and LBVC to LOD in box search (ablation) | GCFS · NuScenes->FS-KITTI complex-structure objects | mAP: 4.65 -> 6.74 | Improved |
| add LFVD and LBVC to LOD in box search (SS) (ablation) | GCFS · NuScenes->FS-KITTI simple-structure objects | mAP: 12.56 -> 13.58 | Improved |
| add MAML-based training (ablation) | GCFS · NuScenes->FS-KITTI 5-shot | AP: 12.35 -> 13.55 | Improved |
| use physical-aware box search instead of FnP (ablation) | GCFS · NuScenes->FS-KITTI 5-shot | AP: 12.58 -> 13.55 | Improved |
| increase shots K from 1 to 40 (ablation) | GCFS · NuScenes->FS-KITTI | overall mAP: 3.44 -> 26.05 | Improved |
| GCFS vs Target-FT in low-shift Waymo->FS-KITTI, common classes | GCFS · Waymo->FS-KITTI common mAP | mAP: 23.06 -> 25.40 (Source-only 26.85) | Mixed |
| swap VoxelRCNN for PV-RCNN backbone (ablation) | GCFS · NuScenes->FS-KITTI | overall mAP: 10.20 -> 14.09 (vs Target-FT) | Improved |
| existing 3D-FSL/GFSL methods vs plain fine-tuning | Proto-Vote, CP-Vote, GFS-Det · NuScenes->FS-KITTI | overall mAP: Target-FT 8.61; CP-Vote 9.06; GFS-Det 6.17 | Mixed |
Closest papers by meaning
- CMDA: Cross-Modal and Domain Adversarial Adaptation for LiDAR-Based 3D Object Detection (AAAI 2024 · similarity 0.89)
- GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detection (CVPR 2025 · similarity 0.89)
- Generalized Few-shot 3D Point Cloud Segmentation with Vision-Language Model (CVPR 2025 · similarity 0.88)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · similarity 0.88)
- Generalizable Multi-Camera 3D Object Detection from a Single Source via Fourier Cross-View Learning (ICML 2025 · similarity 0.88)
- 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.
