Research map / 3D vision / Face and shape
Warp and Learn: Novel Views Generation for Vehicles and Other Objects
Andrea Palazzi, Luca Bergamini, Simone Calderara, Rita Cucchiara
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 synthetic ShapeNet renders to training (real+synth) (ablation) | ICN on Pascal3D+ car · Pascal3D+ car | FID: 77.0 -> 74.1 | Improved |
| add synthetic renders to training (real+synth) (ablation) | ICN on Pascal3D+ chair · Pascal3D+ chair | FID: 88.4 -> 82.8 | Improved |
| add synthetic data at the 180 degree azimuth (ablation) | ICN on Pascal3D+ car · Pascal3D+ car 180 deg | FID: 202.5 -> 168.8 | Improved |
| pre-train on synthetic then fine-tune on real instead of mixing (ablation) | ICN · Pascal3D+ car | visual quality | Hurt |
| add dewarping trick (train from dewarped patches) (ablation) | ICN · Pascal3D+ test | qualitative | Improved |
| add central-crop appearance prior (ablation) | ICN · Pascal3D+ test | qualitative | Improved |
| add adversarial loss (ablation) | ICN · Pascal3D+ test | qualitative | Improved |
| use 2.5D normal sketch instead of silhouette only (ablation) | ICN · Pascal3D+ test | qualitative | Improved |
| use VON fine-tuned on Pascal3D+ instead of original VON | VON · Pascal3D+ car | FID: 126.4 -> 133.2 | Hurt |
| compare ours with VUnet in human A/B test on textured cars | ours · car textured | percent preferring ours: 85.0 | Improved |
| compare ours with VON FT in human A/B test | ours · car avg | percent preferring ours: 97.0 | Improved |
Closest papers by meaning
- AUTO3D: Novel view synthesis through unsupervisely learned variational viewpoint and global 3D representation (ECCV 2020 · similarity 0.88)
- Single-View View Synthesis With Multiplane Images (CVPR 2020 · similarity 0.88)
- Pixel2Mesh++: 3D Mesh Generation and Refinement From Multi-View Images (TPAMI 2022 · similarity 0.88)
- Novel Views of Objects from a Single Image (TPAMI 2016 · similarity 0.87)
- Generative View Synthesis: From Single-view Semantics to Novel-view Images (NeurIPS 2020 · similarity 0.87)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · 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.
