Research map / 3D vision / Face and shape
Shape Retrieval of Non-rigid 3D Human Models
David Pickup, Xianfang Sun, Paul L. Rosin, R. R. Martin, Zhi‐Quan Cheng, Zhouhui Lian, Masaki Aono, A. Ben Hamza, Alex M. Bronstein, Michael M. Bronstein, Shuhui Bu, Umberto Castellani and 19 more
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 |
|---|---|---|---|
| train on the provided training set (APT-trained vs APT) | APT · Real (40 classes) | NN: 0.830 -> 0.910 | Improved |
| add training (supDL vs UnSup32) | Litman dictionary learning · Real (40 classes) | NN: 0.583 -> 0.775 | Improved |
| train on the provided training set (APT-trained vs APT) | APT · FAUST (10 classes) | NN: 0.960 -> 0.990 | Improved |
| use scale-invariant variant (R-BiHDM-s vs R-BiHDM) | R-BiHDM · Real (40 classes) | NN: 0.275 -> 0.720 | Improved |
| reduce Real dataset from 40 to 10 classes | APT-trained · Real (10-class subsets, avg of 100) | NN: 0.910 -> 0.968 | Improved |
| reduce Real dataset from 40 to 10 classes | HKS · Real (10-class subsets, avg of 100) | NN: 0.245 -> 0.625 | Improved |
| remove child models from Synthetic dataset | APT · Synthetic | NN: 0.970 -> 0.955 | Hurt |
| remove child models from Synthetic dataset | supDL · Synthetic | NN: 0.963 -> 0.945 | Hurt |
| evaluate on Synthetic instead of Real data | Spectral Geom. · Spectral Geom. across datasets | NN: 0.313 -> 0.993 | Improved |
| evaluate on Synthetic instead of Real data | WKS · WKS across datasets | NN: 0.326 -> 0.810 | Improved |
| use simple surface area descriptor instead of sophisticated methods | Surface area · Synthetic | 2-T: 0.987 | Improved |
| correlate NN performance with pose-caused errors | all methods · Real / Synthetic / FAUST | correlation: -0.25, -0.50, 0.46 | Mixed |
Closest papers by meaning
- Shape Retrieval of Non-Rigid 3D Human Models | NIST (IJCV 2016 · similarity 0.90)
- Benchmarking and Analyzing 3D Human Pose and Shape Estimation Beyond Algorithms (NeurIPS 2022 · similarity 0.89)
- 3D Shape Sequence of Human Comparison and Classification Using Current and Varifolds (ECCV 2022 · similarity 0.89)
- Accurate 3D Body Shape Regression Using Metric and Semantic Attributes (CVPR 2022 · similarity 0.87)
- End-to-End Recovery of Human Shape and Pose (CVPR 2018 · similarity 0.87)
- Evaluating Multiview Object Consistency in Humans and Image Models (NeurIPS 2024 · similarity 0.86)
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.
