Research map / 3D vision / Face and shape
Few-Shot Single-View 3-D Object Reconstruction with Compositional Priors
Mateusz Michalkiewicz, Sarah Parisot, Stavros Tsogkas, Mahsa Baktashmotlagh, Anders Eriksson, Eugene Belilovsky
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 a global class embedding (GCE) to the Wallace baseline | Wallace · novel classes 1-shot | mean relative IoU gain: 40.2% -> 44.7% | Improved |
| add compositional class embeddings (CGCE) to GCE | GCE · novel classes 1-shot | mean relative IoU gain: 44.7% -> 53.7% | Improved |
| add multi-scale conditional batchnorm embeddings (MCCE) to GCE | GCE · novel classes 1-shot | mean relative IoU gain: 44.7% -> 52.2% | Improved |
| use CGCE instead of Wallace with 10 shots | CGCE · novel classes 10-shot | mean relative IoU gain: 46.5% -> 58.3% | Improved |
| use CGCE instead of Wallace with 25 shots | CGCE · novel classes 25-shot | mean relative IoU gain: 51.9% -> 69.8% | Improved |
| increase shots from 1 to 25 for CGCE | CGCE · novel classes | mean relative IoU gain: 53.7% -> 69.8% | Improved |
| increase shots from 1 to 25 for Wallace | Wallace · novel classes | mean relative IoU gain: 40.2% -> 51.9% | Improved |
| use the correct class embedding rather than a random one at test time (ablation) | GCE · base class plane | IoU: 0.198 -> 0.582 | Improved |
| use the correct class embedding rather than a random one at test time (ablation) | GCE · base class car | IoU: 0.412 -> 0.837 | Improved |
| CGCE vs Wallace on base classes | CGCE · base classes | mean IoU: 0.62 -> 0.62 | No clear effect |
Closest papers by meaning
- Few-Shot Generalization for Single-Image 3D Reconstruction via Priors (ICCV 2019 · similarity 0.92)
- Few-Shot Single-View 3D Reconstruction with Memory Prior Contrastive Network (ECCV 2022 · similarity 0.90)
- Fostering Generalization in Single-View 3D Reconstruction by Learning a Hierarchy of Local and Global Shape Priors (CVPR 2021 · similarity 0.90)
- What Do Single-View 3D Reconstruction Networks Learn? (CVPR 2019 · similarity 0.90)
- Learning to Reconstruct Shapes from Unseen Classes (NeurIPS 2018 · similarity 0.90)
- Learning Shape Priors for Single-View 3D Completion and Reconstruction (ECCV 2018 · similarity 0.90)
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.
