Research map / 3D vision / Face and shape
3D Shape Segmentation With Projective Convolutional Networks
Evangelos Kalogerakis, Melinos Averkiou, Subhransu Maji, Siddhartha Chaudhuri
Check 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 ShapePFCN instead of ShapeBoost | 3D shape part segmentation · ShapeNetCore | Dataset Avg. labeling accuracy: 81.2 -> 87.5 | Improved |
| use ShapePFCN instead of ShapeBoost | 3D shape part segmentation · PSB & COSEG | Dataset Avg. labeling accuracy: 84.2 -> 92.2 | Improved |
| use adaptive coverage-based viewpoints instead of fixed dodecahedron viewpoints (ablation) | ShapePFCN · ShapeNetCore | Dataset Avg. labeling accuracy: 86.2 -> 87.5 | Improved |
| train the FCN and CRF jointly instead of separately (ablation) | ShapePFCN · ShapeNetCore | Dataset Avg. labeling accuracy: 85.9 -> 87.5 | Improved |
| add the surface CRF pairwise terms rather than using only the unary term (ablation) | ShapePFCN · ShapeNetCore | Dataset Avg. labeling accuracy: 82.1 -> 87.5 | Improved |
| initialize the FCN from pretrained VGG instead of training from scratch (ablation) | ShapePFCN · ShapeNetCore | Dataset Avg. labeling accuracy: 85.7 -> 87.5 | Improved |
| add a rendered height-from-ground-plane input channel (ablation) | ShapePFCN with consistently upright shapes · ShapeNetCore, upright | Dataset Avg. labeling accuracy: 87.5 -> 88.8 | Improved |
| use depth images alone instead of shaded and depth images (ablation) | ShapePFCN · ShapeNetCore | labeling accuracy: 2.5% | Hurt |
| use average rather than maximum to aggregate projected view confidences (ablation) | ShapePFCN projection layer · ShapeNetCore | labeling accuracy: about 1% | Hurt |
Closest papers by meaning
- Directionally Convolutional Networks for 3D Shape Segmentation (ICCV 2017 · similarity 0.90)
- PartNet: A Recursive Part Decomposition Network for Fine-Grained and Hierarchical Shape Segmentation (CVPR 2019 · similarity 0.89)
- PartNet: A Large-Scale Benchmark for Fine-Grained and Hierarchical Part-Level 3D Object Understanding (CVPR 2019 · similarity 0.88)
- Learning Fine-Grained Segmentation of 3D Shapes Without Part Labels (CVPR 2021 · similarity 0.87)
- JSIS3D: Joint Semantic-Instance Segmentation of 3D Point Clouds With Multi-Task Pointwise Networks and Multi-Value Conditional Random Fields (CVPR 2019 · similarity 0.87)
- Semantic Segmentation With Boundary Neural Fields (CVPR 2016 · 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.
