Research map / Probabilistic and scientific modeling / Variational and generative sampling
Localization with Sampling-Argmax
Jiefeng Li, Tong Chen, Ruiqi Shi, Yujing Lou, Yong-Lu Li, Cewu Lu
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 soft-argmax with sampling-argmax (triangular basis) (ablation) | SimplePose · COCO Keypoint | mAP: 64.5 -> 69.8 | Improved |
| add variance regularization to soft-argmax (ablation) | soft-argmax · COCO Keypoint | mAP: 64.5 -> 60.6 | Hurt |
| add distribution regularization to soft-argmax (ablation) | soft-argmax · COCO Keypoint | mAP: 64.5 -> 55.6 | Hurt |
| increase number of samples from 1 to 50 (ablation) | sampling-argmax triangular · COCO Keypoint | mAP: 69.7 -> 69.8 | No clear effect |
| increase number of samples from 1 to 30 (ablation) | sampling-argmax uniform · COCO Keypoint | mAP: 67.8 -> 68.2 | No clear effect |
| use sampling-argmax triangular instead of soft-argmax | SimplePose · COCO Keypoint correlation with correctness | Pearson correlation: 0.233 -> 0.432 | Improved |
| replace soft-argmax with sampling-argmax triangular (ablation) | 3D pose model · Human3.6M | MPJPE: 50.4 -> 49.5 | Improved |
| add variance regularization to soft-argmax (ablation) | soft-argmax · Human3.6M | MPJPE: 50.4 -> 49.7 | Improved |
| add distribution regularization to soft-argmax (ablation) | soft-argmax · Human3.6M | MPJPE: 50.4 -> 51.9 | Hurt |
| use Gaussian sampling basis instead of triangular (ablation) | sampling-argmax · Human3.6M | MPJPE: 49.5 -> 50.9 | Hurt |
| replace soft-argmax with sampling-argmax uniform | retina regression model · MSHC | MAD: 3.08 -> 0.735 | Improved |
| add variance regularization to soft-argmax (ablation) | soft-argmax · MSHC | MAD: 3.08 -> 0.743 | Improved |
Closest papers by meaning
- Stochastic Localization via Iterative Posterior Sampling (ICML 2024 · similarity 0.85)
- Learning Where to Look: Self-supervised Viewpoint Selection for Active Localization using Geometrical Information (ECCV 2024 · similarity 0.84)
- Unsupervised Object Localization with Representer Point Selection (ICCV 2023 · similarity 0.82)
- CPR++: Object Localization via Single Coarse Point Supervision (TPAMI 2024 · similarity 0.82)
- Active Neural Localization (ICLR 2018 · similarity 0.82)
- From Variance to Veracity: Unbundling and Mitigating Gradient Variance in Differentiable Bundle Adjustment Layers (CVPR 2024 · similarity 0.80)
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.
