Research map / Detection and segmentation / 3D detection and driving
3D Part Guided Image Editing for Fine-Grained Object Understanding
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 |
|---|---|---|---|
| train on 3D-part-guided edited images instead of existing datasets | Mask-RCNN · CUS Dataset | 2D Detection (IoU): 0.751 -> 0.775 | Helped |
| train on 3D-part-guided edited images instead of existing datasets | Mask-RCNN · CUS Dataset | Ins. Seg. (IoU): 0.704 -> 0.721 | Helped |
| use the two-backbone network with edited images rather than the strongest listed baseline | CUS Dataset · CUS Dataset | 2D Detection (IoU): 0.775 -> 0.862 | Helped |
| use the two-backbone network with edited images rather than the strongest listed baseline | CUS Dataset · CUS Dataset | Ins. Seg. (IoU): 0.721 -> 0.815 | Helped |
| freeze the single backbone rather than re-train it on edited data (ablation) | CUS Dataset · CUS Dataset | 2D Detection (mAP): 0.136 -> 0.672 | Helped |
| use two frozen backbones specialized for car and general features instead of one frozen backbone (ablation) | CUS Dataset · CUS Dataset | Part Seg. (mAP): 0.273 -> 0.314 | Helped |
| use two frozen backbones specialized for car and general features instead of one frozen backbone (ablation) | CUS Dataset · CUS Dataset | State Description: 0.837 -> 0.874 | Helped |
| add an environment map when rendering reverse sides of parts (ablation) | Two-Backbone Network · CUS Dataset | Part Seg. (mAP): 0.221 -> 0.314 | Helped |
| add an environment map when rendering reverse sides of parts (ablation) | Two-Backbone Network · CUS Dataset | 2D Detection (mAP): 0.688 -> 0.701 | Helped |
| increase the amount of edited training data (ablation) | Two-Backbone Network · CUS Dataset | network performance | Mixed |
Closest papers by meaning
- OpenBox: Annotate Any Bounding Boxes in 3D (NeurIPS 2025 · similarity 0.87)
- DriveEditor: A Unified 3D Information-Guided Framework for Controllable Object Editing in Driving Scenes (AAAI 2025 · similarity 0.86)
- SceneCrafter: Controllable Multi-View Driving Scene Editing (CVPR 2025 · similarity 0.85)
- DriveGEN: Generalized and Robust 3D Detection in Driving via Controllable Text-to-Image Diffusion Generation (CVPR 2025 · similarity 0.85)
- PartNeXt: A Next-Generation Dataset for Fine-Grained and Hierarchical 3D Part Understanding (NeurIPS 2025 · similarity 0.85)
- CorrBEV: Multi-View 3D Object Detection by Correlation Learning with Multi-modal Prototypes (CVPR 2025 · similarity 0.85)
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.