Research map / 3D vision / Face and shape
View Inter-Prediction GAN: Unsupervised Representation Learning for 3D Shapes by Learning Global Shape Memories to Support Local View Predictions
Read the paperCheck 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 |
|---|---|---|---|
| Use view inter-prediction with a shared trainable global memory (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 84.58 -> 94.05 | Helped |
| Add decoder R and neighbor-feature prediction (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 75.77 -> 94.05 | Helped |
| Add adversarial loss with beta=0.05 (ablation) | VIP-GAN with alpha=3 · ModelNet10 | instance accuracy: 92.51 -> 94.05 | Helped |
| Set adversarial-loss weight beta to 0.1 (ablation) | VIP-GAN with alpha=3 · ModelNet10 | instance accuracy: 94.05 -> 92.84 | Hurt |
| Set adversarial-loss weight beta to 0.01 (ablation) | VIP-GAN with alpha=3 · ModelNet10 | instance accuracy: 94.05 -> 91.19 | Hurt |
| Set neighbor-loss weight alpha to 1 instead of 3 (ablation) | VIP-GAN with beta=0.05 · ModelNet10 | instance accuracy: 94.05 -> 92.73 | Hurt |
| Set neighbor-loss weight alpha to 5 instead of 3 (ablation) | VIP-GAN with beta=0.05 · ModelNet10 | instance accuracy: 94.05 -> 93.50 | Hurt |
| Replace the simple deconvolutional network with a more complex one (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 84.80 -> 75.77 | Hurt |
| Increase feature dimension from 1024 to 4096 (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 92.29 -> 94.05 | Helped |
| Use 4 neighbors per section rather than 2 or 6 (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 93.17 / 93.50 -> 94.05 | Helped |
| Use the full 12-view sequence instead of 6 or 3 views (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 92.62 / 92.51 -> 94.05 | Helped |
| Replace GAN with conditional GAN (ablation) | VIP-GAN · ModelNet10 | instance accuracy: 94.05 -> 89.10 | Hurt |
Closest papers by meaning
- Do 2D GANs Know 3D Shape? Unsupervised 3D Shape Reconstruction from 2D Image GANs (ICLR 2021 · similarity 0.88)
- Unsupervised 3D Shape Completion Through GAN Inversion (CVPR 2021 · similarity 0.87)
- Multiview Aggregation for Learning Category-Specific Shape Reconstruction (NeurIPS 2019 · similarity 0.87)
- GVCNN: Group-View Convolutional Neural Networks for 3D Shape Recognition (CVPR 2018 · similarity 0.86)
- Pixels, Voxels, and Views: A Study of Shape Representations for Single View 3D Object Shape Prediction (CVPR 2018 · similarity 0.86)
- Duoduo CLIP: Efficient 3D Understanding with Multi-View Images (ICLR 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.