Research map / 3D vision / 3D generation
Triplet-Center Loss for Multi-View 3D Object Retrieval
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 |
|---|---|---|---|
| Add TCL to softmax training (ablation) | MVCNN · ModelNet40 | mAP: 80.2 -> 88.0 | Helped |
| Replace TCL with center loss while retaining softmax (ablation) | MVCNN · ModelNet40 | mAP: 88.0 -> 83.5 | Hurt |
| Replace TCL with triplet loss (ablation) | MVCNN · ModelNet40 | mAP: 88.0 -> 80.0 | Hurt |
| Add TCL without softmax loss (ablation) | MVCNN · ModelNet40 | mAP: 80.2 -> 86.7 | Helped |
| Add softmax loss to TCL (ablation) | MVCNN · ModelNet40 | mAP: 86.7 -> 88.0 | Helped |
| Add TCL to softmax training (ablation) | VoxNet · ModelNet40 | mAP: 70.0 -> 73.2 | Helped |
| Add TCL to softmax training (ablation) | PointNet · ModelNet40 | mAP: 70.5 -> 74.5 | Helped |
| Set TCL weight λ to 0 instead of combining TCL with softmax (ablation) | MVCNN · ModelNet40 | mAP: nearly 80.0% -> 88.0% | Hurt |
| Vary λ from 0.01 to 10 (ablation) | TCL + softmax loss · ModelNet40 | mAP | No clear effect |
| Use a margin m other than 5 (ablation) | TCL + softmax loss · ModelNet40 | mAP: 88.0% | Hurt |
Closest papers by meaning
- Angular Triplet-Center Loss for Multi-View 3D Shape Retrieval (AAAI 2019 · similarity 0.93)
- Learning With Batch-Wise Optimal Transport Loss for 3D Shape Recognition (CVPR 2019 · similarity 0.87)
- A Theoretically Sound Upper Bound on the Triplet Loss for Improving the Efficiency of Deep Distance Metric Learning (CVPR 2019 · similarity 0.87)
- Improved Deep Metric Learning with Multi-class N-pair Loss Objective (NeurIPS 2016 · similarity 0.86)
- Deep Correlated Metric Learning for Sketch-based 3D Shape Retrieval (AAAI 2017 · similarity 0.86)
- Collapse-Aware Triplet Decoupling for Adversarially Robust Image Retrieval (ICML 2024 · similarity 0.83)
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.