Research map / 3D vision / Human pose estimation
Learning in an Uncertain World: Representing Ambiguity Through Multiple Hypotheses
Christian Rupprecht, Iro Laina, Robert DiPietro, Maximilian Baust, Federico Tombari, Nassir Navab, Gregory D. Hager
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 |
|---|---|---|---|
| increase hypotheses from 1 (SHP) to 10 (10-MHP) (ablation) | pose regression CNN · LSP pose estimation (oracle) | strict PCP: 59.7 -> 62.8 | Improved |
| replace single prediction with 10 hypotheses (ablation) | future frame CNN · NTU RGB-D put on a hat/cap | sharpness: 319.5 -> 419.7 | Improved |
| replace single prediction with 10 hypotheses (ablation) | future frame CNN · NTU RGB-D put on a hat/cap | min MSE: 960.6 -> 728.5 | Improved |
| replace single prediction with 9 hypotheses (ablation) | ResNet-101 · VOC07 test | mAP: 83.8 -> 85.1 | Improved |
| replace single prediction with 9 hypotheses (ablation) | ResNet-101 · VOC12 val | mAP: 86.9 -> 87.6 | Improved |
| replace single prediction with 5 hypotheses (ablation) | ResNet-101 · COCO val | mAP: 65.2 -> 67.8 | Improved |
| replace single prediction with 5 hypotheses (ablation) | ResNet-101 · COCO val | mAP@10: 81.0 -> 83.3 | Improved |
| increase hypotheses from 9 to 13 (ablation) | ResNet-101 MHP · VOC12 val | mAP: 87.6 -> 87.0 | Hurt |
| increase hypotheses from 5 to 13 (ablation) | ResNet-101 MHP · VOC07 test | mAP: 84.7 -> 84.7; 9-MHP peaks at 85.1 | Mixed |
| use shared-architecture 4-MHP FCN8s instead of MCL ensemble | FCN8s · VOC2012 segmentation | mean IoU: 69.1 -> 70.3 | Improved |
| use ResNet-101 instead of ResNet-50 or VGG-16 backbone (ablation) | MHP classifier · VOC multi-label | mAP: 2-3% worse for ResNet-50 and VGG-16 | Improved |
Closest papers by meaning
- Plausible Uncertainties for Human Pose Regression (ICCV 2023 · similarity 0.89)
- Benchmarking Uncertainty Disentanglement: Specialized Uncertainties for Specialized Tasks (NeurIPS 2024 · similarity 0.87)
- From Risk to Uncertainty: Generating Predictive Uncertainty Measures via Bayesian Estimation (ICLR 2025 · similarity 0.86)
- ValUES: A Framework for Systematic Validation of Uncertainty Estimation in Semantic Segmentation (ICLR 2024 · similarity 0.85)
- Learning from Uncertain Data: From Possible Worlds to Possible Models (NeurIPS 2024 · similarity 0.85)
- Experts Don't Cheat: Learning What You Don't Know By Predicting Pairs (ICML 2024 · 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.
