Research map / 3D vision / Face and shape
Enhancing 2D Representation via Adjacent Views for 3D Shape Retrieval
Cheng Xu, Zhaoqun Li, Qiang Qiu, Biao Leng, Jingfei Jiang
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 center loss to softmax loss (ablation) | NCENet Parallel · ModelNet40 | mAP: 83.47% -> 86.27% | Improved |
| add center loss to softmax loss (ablation) | NCENet Serial · ModelNet40 | mAP: 83.04% -> 87.05% | Improved |
| use correlation unit instead of max pooling of adjacent features (ablation) | NCENet Parallel · ModelNet40 | mAP: 84.12% -> 86.27% | Improved |
| use correlation unit instead of average pooling of adjacent features (ablation) | NCENet Serial · ModelNet40 | mAP: 81.29% -> 87.05% | Improved |
| use serial instead of parallel enhanced structure (ablation) | NCENet · ModelNet40 | mAP: 86.27% -> 87.05% | Improved |
| use serial instead of parallel enhanced structure (ablation) | NCENet · ShapeNetCore55 perturbed (micro) | mAP: 89.6 -> 89.0 | Hurt |
| increase number of center views from 1 to 4 at test (ablation) | Our-Parallel · ModelNet40 | mAP: 0.819 -> 0.867 | Improved |
| increase number of center views from 6 to 10 (ablation) | Our-Serial · ModelNet40 | mAP: 0.871 -> 0.868 | No clear effect |
| increase number of neighbor views from 1 to 8 (ablation) | Our-Serial · ModelNet40 | mAP: 0.853 -> 0.871 | Improved |
| increase number of neighbor views from 1 to 8 (ablation) | Our-Parallel · ModelNet40 | mAP: 0.852 -> 0.863 | Improved |
| replace MVCNN max pooling with Our-Serial enhancement on same GoogLeNet-bn (ablation) | NCENet · ModelNet40 | mAP: 85.49% -> 87.05% | Improved |
| occlude all neighboring views to 80% level | Our-Parallel · ModelNet40 | mAP: decreases slightly by 2.55% | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GIFT — GIFT: A Real-Time and Scalable 3D Shape Search Engine (CVPR 2016) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Emphasizing 3D Properties in Recurrent Multi-View Aggregation for 3D Shape Retrieval (AAAI 2018 · similarity 0.88)
- GVCNN: Group-View Convolutional Neural Networks for 3D Shape Recognition (CVPR 2018 · similarity 0.88)
- MLVCNN: Multi-Loop-View Convolutional Neural Network for 3D Shape Retrieval (AAAI 2019 · similarity 0.88)
- View-GCN: View-Based Graph Convolutional Network for 3D Shape Analysis (CVPR 2020 · similarity 0.88)
- Single Image 3D Shape Retrieval via Cross-Modal Instance and Category Contrastive Learning (ICCV 2021 · similarity 0.88)
- Learning View-Based Graph Convolutional Network for Multi-View 3D Shape Analysis (TPAMI 2022 · 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.
