Research map / 3D vision / Human pose estimation
Graph Stacked Hourglass Networks for 3D Human Pose Estimation
Check 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 |
|---|---|---|---|
| replace SeqRes with GraphSH while using PreAggr | PreAggr-based 2D-to-3D pose estimation · Human3.6M | MPJPE (mm): 37.8 -> 35.8 | Helped |
| replace SeqRes with GraphSH while using SemGConv | SemGConv-based 2D-to-3D pose estimation · Human3.6M | MPJPE (mm): 52.5 -> 39.2 | Helped |
| replace SeqRes with GraphSH while using Vanilla GCN | Vanilla GCN-based 2D-to-3D pose estimation · Human3.6M | MPJPE (mm): 94.4 -> 59.1 | Helped |
| replace input-dependent gPool/gUnpool with fixed skeletal pooling and unpooling (ablation) | GraphSH · Human3.6M, ground-truth 2D keypoints | MPJPE (mm): 40.6 -> 35.8 | Helped |
| replace SAGPool/gUnpool with fixed skeletal pooling and unpooling (ablation) | GraphSH · Human3.6M, ground-truth 2D keypoints | MPJPE (mm): 41.5 -> 35.8 | Helped |
| add skeletal pooling and unpooling to replace processing only at the highest scale (ablation) | GraphSH · Human3.6M, ground-truth 2D keypoints | MPJPE (mm): 38.2 -> 35.8 | Helped |
| add multi-level intermediate features (ablation) | GraphSH · Human3.6M, ground-truth 2D keypoints | MPJPE (mm): 38.0 -> 35.8 | Helped |
| add an SE block to weight concatenated multi-level features (ablation) | GraphSH with multi-level features · Human3.6M, ground-truth 2D keypoints | MPJPE (mm): 37.1 -> 35.8 | Helped |
Closest papers by meaning
- Graph and Temporal Convolutional Networks for 3D Multi-person Pose Estimation in Monocular Videos (AAAI 2021 · similarity 0.89)
- Multi-Context Attention for Human Pose Estimation (CVPR 2017 · similarity 0.88)
- Semantic Graph Convolutional Networks for 3D Human Pose Regression (CVPR 2019 · similarity 0.87)
- Deep Semantic Graph Transformer for Multi-View 3D Human Pose Estimation (AAAI 2024 · similarity 0.87)
- Thin-Slicing Network: A Deep Structured Model for Pose Estimation in Videos (CVPR 2017 · similarity 0.86)
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation (CVPR 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.