Research map / 3D vision / Human pose estimation
DGCN: Dynamic Graph Convolutional Network for Efficient Multi-Person Pose Estimation
Read the paperCheck 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 |
|---|---|---|---|
| add simple GCN head (identity adjacency) to baseline (ablation) | Pifpaf ResNet-50 · COCO val2017 | AP: 0.626 -> 0.636 | Helped |
| use fixed hand-crafted skeleton adjacency instead of identity (ablation) | GCN head · COCO val2017 | AP: 0.636 -> 0.639 | Helped |
| use soft distance-based adjacency instead of hand-crafted (ablation) | GCN head · COCO val2017 | AP: 0.639 -> 0.641 | Helped |
| sample dynamic Bernoulli adjacency instead of fixed soft adjacency (ablation) | GCN head · COCO val2017 | AP: 0.641 -> 0.646 | Helped |
| add DGCN head on F8 (pyramid) (ablation) | DGCN-50 · COCO val2017 | AP: 0.646 -> 0.651 | Helped |
| add third DGCN head on F4 (ablation) | DGCN-50 · COCO val2017 | AP: 0.651 -> 0.652 | No clear effect |
| vary adjacency scale gamma from 0.01 to 100 (ablation) | DGCN soft adjacency · COCO val2017 | AP: 0.638 to 0.641 | No clear effect |
| add one DGCN head to Pifpaf | Pifpaf ResNet-101 · COCO val2017 | AP: 0.657 -> 0.673 | Helped |
| add DGCN head to Pifpaf on CrowdPose | Pifpaf ResNet-50 · CrowdPose | AP: 0.563 -> 0.591 | Helped |
| add DGCN head to Pifpaf with DenseNet121 | Pifpaf DenseNet121 · COCO val | AP: 0.618 -> 0.636 | Helped |
Closest papers by meaning
- Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation (AAAI 2020 · similarity 0.89)
- Towards Accurate Multi-Person Pose Estimation in the Wild (CVPR 2017 · similarity 0.89)
- Cascaded Pyramid Network for Multi-Person Pose Estimation (CVPR 2018 · similarity 0.89)
- Graph and Temporal Convolutional Networks for 3D Multi-person Pose Estimation in Monocular Videos (AAAI 2021 · similarity 0.88)
- Deep Dual Consecutive Network for Human Pose Estimation (CVPR 2021 · similarity 0.88)
- Learning Dynamics via Graph Neural Networks for Human Pose Estimation and Tracking (CVPR 2021 · similarity 0.87)
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.