Research map / 3D vision / Human pose estimation
PoseImageNet: Pose Estimation for Extensive Classes Based on Rich Structure Prototypes
Junjie Chen, Hong Cao, Weixiang Tao, Yuming Fang, Jiebin Yan, Yifan Zuo
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 |
|---|---|---|---|
| add adaptive matching of shared queries to targets (instead of fixed 1-to-1 mapping) (ablation) | Ours · PoseImageNet test (all classes) | mAP@0.2: 35.3 -> 57.0 | Improved |
| add adaptive matching of shared queries to targets (instead of fixed 1-to-1 mapping) (ablation) | Ours · PoseImageNet test (all classes) | AVG mAP: 22.0 -> 44.8 | Improved |
| use ground-truth class label instead of predicted class to fetch the cost matrix (ablation) | Ours · PoseImageNet test (all classes) | mAP@0.2: 57.0 -> 79.5 | Improved |
| use ground-truth class label instead of predicted class to fetch the cost matrix (ablation) | Ours · PoseImageNet test (all classes) | AVG mAP: 44.8 -> 63.3 | Improved |
| Ours (query-based model with matching supervision) versus RTMO | multi-class pose baseline · PoseImageNet test (all classes) | mAP@0.2: 48.2 -> 57.0 | Improved |
| Ours versus GroupPose on wear items super-class | multi-class pose baseline · PoseImageNet test (wear items super-class) | mAP@0.2: 46.2 -> 62.3 | Improved |
| Ours versus RTMO at fine threshold | multi-class pose baseline · PoseImageNet test (all classes) | mAP@0.05: 17.5 -> 27.2 | Improved |
| increase classifier depth from ResNet-18 to ResNet-101 | structure prototype classification · PoseImageNet prototype classification | Top-1 accuracy: 59.8 -> 66.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GroupPose — Group Pose: A Simple Baseline for End-to-End Multi-Person Pose Estimation (ICCV 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Deep Learning-Based Object Pose Estimation: A Comprehensive Survey (IJCV 2026 · similarity 0.88)
- DSC-PoseNet: Learning 6DoF Object Pose Estimation via Dual-Scale Consistency (CVPR 2021 · similarity 0.88)
- Learning Temporal Pose Estimation from Sparsely-Labeled Videos (NeurIPS 2019 · similarity 0.88)
- Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation (ICCV 2017 · similarity 0.88)
- Pose Prior Learner: Unsupervised Categorical Prior Learning for Pose Estimation (ICLR 2026 · similarity 0.88)
- PriorPose: Reference-Guided Joint Deformation and Alignment for Category-Level Object Pose Estimation (ECCV 2026 · similarity 0.88)
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.
