Research map / 3D vision / Face and shape
Geo-SIC: Learning Deformable Geometric Shapes in Deep Image Classifiers
Jian Wang, Miaomiao Zhang
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 |
|---|---|---|---|
| joint learning of shape and classifier instead of two-step preprocessing (ResNet18) (ablation) | Geo-SIC:Res · 2D QuickDraw | Accuracy: 0.911 -> 0.935 | Improved |
| joint learning instead of two-step (AlexNet) (ablation) | Geo-SIC:Alex · 2D QuickDraw | Accuracy: 0.891 -> 0.928 | Improved |
| joint learning instead of two-step (CNN) (ablation) | Geo-SIC:CNN · 2D QuickDraw | Accuracy: 0.869 -> 0.897 | Improved |
| joint learning instead of two-step (VGG19) (ablation) | Geo-SIC:VGG · 2D QuickDraw | Accuracy: 0.895 -> 0.927 | Improved |
| add learned geometric shape features to ResNet18 | ResNet18 · 2D QuickDraw | Accuracy: 0.875 -> 0.935 | Improved |
| add learned geometric shape features to AlexNet | AlexNet · 2D QuickDraw | Accuracy: 0.880 -> 0.928 | Improved |
| add learned geometric shape features to CNN | 3D CNN · 2D QuickDraw | Accuracy: 0.861 -> 0.897 | Improved |
| add learned geometric shape features to VGG19 | VGG19 · 2D QuickDraw | Accuracy: 0.883 -> 0.927 | Improved |
| joint learning instead of two-step (ResNet18) (ablation) | Geo-SIC:Res · OASIS 3D brain MRI | Accuracy: 0.833 -> 0.873 | Improved |
| joint learning instead of two-step (VGG19) (ablation) | Geo-SIC:VGG · OASIS 3D brain MRI | Accuracy: 0.865 -> 0.852 | Hurt |
| joint learning instead of two-step (AlexNet) (ablation) | Geo-SIC:Alex · OASIS 3D brain MRI | Accuracy: 0.800 -> 0.835 | Improved |
| add learned geometric shape features to ResNet18 | ResNet18 · OASIS 3D brain MRI | Accuracy: 0.805 -> 0.873 | Improved |
Closest papers by meaning
- Neural Convolutional Surfaces (CVPR 2022 · similarity 0.87)
- FeaStNet: Feature-Steered Graph Convolutions for 3D Shape Analysis (CVPR 2018 · similarity 0.86)
- Linear Shape Deformation Models With Local Support Using Graph-Based Structured Matrix Factorisation (CVPR 2016 · similarity 0.86)
- Survey and Evaluation of Neural 3D Shape Classification Approaches (TPAMI 2021 · similarity 0.86)
- Joint Learning of 3D Shape Retrieval and Deformation (CVPR 2021 · similarity 0.86)
- 4Deform: Neural Surface Deformation for Robust Shape Interpolation (CVPR 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.
