Research map / Label-efficient and robust learning / Recognition and long-tail
MVP-N: A Dataset and Benchmark for Real-World Multi-View Object Classification
Ren Wang, Jiayue Wang, Tae Sung Kim, JINSUNG KIM, Hyuk-Jae Lee
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 |
|---|---|---|---|
| replace max-pooling aggregation with DAN self-attention | MVCNN-new · MVP-N test | MVA: 89.35 ± 1.21 -> 91.61 ± 0.94 | Improved |
| add GVCNN view grouping | MVCNN-new · MVP-N test | MVA: 89.35 ± 1.21 -> 85.42 ± 1.37 | Hurt |
| add CVR optimal-transport canonical views and transformer | MVCNN-new · MVP-N test | MVA: 89.35 ± 1.21 -> 79.99 ± 2.52 | Hurt |
| use predicted class probabilities as soft labels (SB vs HB) | hard bootstrapping · MVP-N test | MCDU: 0.3899 ± 0.0068 -> 0.2862 ± 0.0107 | Improved |
| use predicted class probabilities as soft labels (SB vs HB) | hard bootstrapping · MVP-N test | MVA: 83.20 ± 1.11 -> 83.85 ± 0.84 | Improved |
| train with HPIQ human information-quantity soft labels instead of CE | cross-entropy · MVP-N test | MVA: 83.37 ± 1.05 -> 94.36 ± 0.56 | Improved |
| train with HPIQ soft labels instead of CE | cross-entropy · MVP-N test | SVA: 78.65 ± 0.44 -> 63.31 ± 0.38 | Hurt |
| SAT instead of SEAL | SEAL · MVP-N test | MVA: 86.42 ± 0.74 -> 87.37 ± 1.15 | Improved |
| label flip components (HB, DHB, LRT, PLC) | CE · MVP-N test | MVA: 83.37 ± 1.05 -> 82.96 ± 1.06 | No clear effect |
| increase uninformative views from 2-view to 6-view case | HPIQ · MVP-N test | MVA: 97.73 ± 0.38 -> 92.07 ± 0.91 | Hurt |
| DAN instead of soft label methods at six views | soft label methods · MVP-N | MVA | Mixed |
Closest papers by meaning
- Multi-View Multi-Label Classification via View-Label Matching Selection (AAAI 2025 · similarity 0.87)
- Noisy Label Calibration for Multi-View Classification (AAAI 2025 · similarity 0.85)
- A Two-Stage Information Extraction Network for Incomplete Multi-View Multi-Label Classification (AAAI 2024 · similarity 0.84)
- Semi-Supervised Multi-View Multi-Label Learning with View-Specific Transformer and Enhanced Pseudo-Label (AAAI 2025 · similarity 0.84)
- View-Category Interactive Sharing Transformer for Incomplete Multi-View Multi-Label Learning (CVPR 2024 · similarity 0.84)
- OpenViewer: Openness-Aware Multi-View Learning (AAAI 2025 · similarity 0.84)
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.
