Research map / Vision-language and multimodal / CLIP and zero-shot
VQA-Diff: Exploiting VQA and Diffusion for Zero-Shot Image-to-3D Vehicle Asset Generation in Autonomous Driving
YIBO LIU, Zheyuan Yang, Guile Wu, Yuan Ren, Kejian Lin, Liu Bingbing, Yang Liu, JINJUN SHAN
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 |
|---|---|---|---|
| use multi-expert DMs instead of one single DM for 16 views (ablation) | VQA-Diff structure generation · BMW X5 structure generation | FID: 192.55 -> 122.91 | Improved |
| train the single DM for 100 epochs instead of 50 (ablation) | Single DM baseline · BMW X5 structure generation | FID: 192.55 -> 189.04 | No clear effect |
| reduce views per single DM image from 16 to 9 (ablation) | Single DM baseline · BMW X5 structure generation | FID: 192.55 -> 150.31 | Improved |
| add ControlNet appearance generation (ablation) | VQA-Diff · Waymo | FID: 186.98 -> 163.40 | Improved |
| add ControlNet to NFI trained on ShapeNet (ablation) | NFI · Waymo | FID: 428.35 -> 267.47 | Improved |
| use VQA-derived prompt for geometry instead of NFI when both trained on ShapeNet (ablation) | VQA-Diff · Waymo | FID: 267.47 -> 186.98 | Improved |
| apply VQA-Diff to teddy bear instead of vehicles | VQA-Diff · Generic objects (qualitative) | structure consistency | Hurt |
Closest papers by meaning
- VGGDrive: Empowering Vision-Language Models with Cross-View Geometric Grounding for Autonomous Driving (CVPR 2026 · similarity 0.88)
- NuScenes-QA: A Multi-Modal Visual Question Answering Benchmark for Autonomous Driving Scenario (AAAI 2024 · similarity 0.87)
- VR-Drive: Viewpoint-Robust End-to-End Driving with Feed-Forward 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.87)
- Structural Information Guided Multimodal Pre-training for Vehicle-Centric Perception (AAAI 2024 · similarity 0.87)
- DriveGEN: Generalized and Robust 3D Detection in Driving via Controllable Text-to-Image Diffusion Generation (CVPR 2025 · similarity 0.87)
- DistillNeRF: Perceiving 3D Scenes from Single-Glance Images by Distilling Neural Fields and Foundation Model Features (NeurIPS 2024 · 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.
