Research map / Detection and segmentation / 3D detection and driving
Detect Anything 3D in the Wild
Hanxue Zhang, Haoran Jiang, Qingsong Yao, Yanan Sun, Renrui Zhang, Hao Zhao, Hongyang Li, Hongzi Zhu, Zetong Yang
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 depth map and camera intrinsic modules (ablation) | DetAny3D · Omni3D (10% data) | AP3D: 5.81 -> 10.10 | Improved |
| merge depth-pretrained DINO with SAM (ablation) | DetAny3D · Omni3D (10% data) | AP3D: 10.10 -> 20.20 | Improved |
| add 2D Aggregator for fusing SAM and DINO features (ablation) | DetAny3D · Omni3D (10% data) | AP3D: 20.20 -> 23.21 | Improved |
| add zero-initialized embedding mixer (ZEM) (ablation) | DetAny3D · Omni3D (10% data) | AP3D: 23.21 -> 25.80 | Improved |
| use pretrained SAM weights (ablation) | DetAny3D · Omni3D | AP3D: 19.12 -> 25.80 | Improved |
| use DINO depth backbone instead of ConvNeXt (ablation) | DetAny3D · Omni3D | AP3D: 23.11 -> 25.80 | Improved |
| add intrinsic prompt (ablation) | DetAny3D · Omni3D box prompt | AP3D: 32.16 -> 34.38 | Improved |
| add DA3D dataset to training (ablation) | DetAny3D · Cityscapes3D novel camera | AP3D (traditional / target-aware): 8.31 / 11.68 -> 11.05 / 15.71 | Improved |
| add DA3D dataset to training (ablation) | DetAny3D · Omni3D in-domain | AP3D: 24.33 -> 24.92 | Improved |
| DetAny3D vs OVMono3D with GT box prompts | OVMono3D · ARKitScenes zero-shot | AP3D: 14.12 -> 57.72 | Improved |
| DetAny3D vs Cube R-CNN with Cube R-CNN prompts | Cube R-CNN · Omni3D_OUT KITTI | AP3D: 36.0 -> 35.8 | No clear effect |
| DetAny3D vs OVMono3D with Cube R-CNN prompts | OVMono3D · Omni3D objectron | AP3D: 58.87 -> 54.42 | Hurt |
Closest papers by meaning
- Uni$^2$Det: Unified and Universal Framework for Prompt-Guided Multi-dataset 3D Detection (ICLR 2025 · similarity 0.90)
- Training an Open-Vocabulary Monocular 3D Detection Model without 3D Data (NeurIPS 2024 · similarity 0.90)
- V-DETR: DETR with Vertex Relative Position Encoding for 3D Object Detection (ICLR 2024 · similarity 0.88)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.88)
- Towards Generalizable Multi-Camera 3D Object Detection via Perspective Rendering (AAAI 2025 · similarity 0.87)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 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.
