Research map / 3D vision / Face and shape
Democratising 2D Sketch to 3D Shape Retrieval Through Pivoting
Pinaki Nath Chowdhury, Ayan Kumar Bhunia, Aneeshan Sain, Subhadeep Koley, Tao Xiang, Yi-Zhe Song
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 pivot-based pretraining (all-shot fine-tuning) (ablation) | FG-SBSR baseline · Fine-grained SBSR chairs | Acc.@1: 10.5 -> 48.7 | Improved |
| add distribution-matching loss Ldist to pivoting (ablation) | pivot-based learning · Fine-grained SBSR chairs (zero-shot) | Acc.@1: 46.5 -> 49.3 | Improved |
| add auxiliary BPnP lifting loss (ablation) | pivoting with Ldist · Fine-grained SBSR chairs (zero-shot) | Acc.@1: 49.3 -> 55.8 | Improved |
| fine-tune zero-shot model on paired sketch-3D shape data (ablation) | full method · Fine-grained SBSR chairs | Acc.@1: 55.8 -> 58.5 | Improved |
| train pivoting on chairs only, evaluate on lamps | zero-shot method · Fine-grained SBSR lamps | Acc.@1: 55.79 (chairs) -> 21.63 (lamps) | Hurt |
| reconstruction-based lifting (SDFSketch/PSGNSketch) on category-level retrieval | 2D sketch encoder · SHREC'14 | mAP: SDFSketch 29.9, PSGNSketch 30.1 vs 2D Enc. up to 30.6 | No clear effect |
| train on synthetic sketches (2D encoder methods) vs 3D encoders | fine-grained SBSR baselines · Fine-grained SBSR chairs | Acc.@1: GANSketch 37.53 vs MVAttn 56.72 | Hurt |
Closest papers by meaning
- Freeview Sketching: View-Aware Fine-Grained Sketch-Based Image Retrieval (ECCV 2024 · similarity 0.90)
- Learning Barycentric Representations of 3D Shapes for Sketch-Based 3D Shape Retrieval (CVPR 2017 · similarity 0.89)
- Deep Sketch-Shape Hashing With Segmented 3D Stochastic Viewing (CVPR 2019 · similarity 0.89)
- SketchSampler: Sketch-Based 3D Reconstruction via View-Dependent Depth Sampling (ECCV 2022 · similarity 0.88)
- Sketch Down the FLOPs: Towards Efficient Networks for Human Sketch (CVPR 2025 · similarity 0.87)
- It's All About Your Sketch: Democratising Sketch Control in Diffusion Models (CVPR 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.
