Research map / 3D vision / Human pose estimation
Learning Temporal Pose Estimation from Sparsely-Labeled Videos
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 |
|---|---|---|---|
| Use PoseWarper instead of FlowNet2 optical-flow propagation | Video pose propagation · PoseTrack2017 val | mAP: 83.8 -> 88.7 | Helped |
| Use PoseWarper instead of the strongest single-frame baseline | Multi-person pose estimation · PoseTrack2017 val | mAP: 80.7 -> 81.2 | Helped |
| Use PoseWarper instead of the strongest listed baseline | Multi-person pose estimation · PoseTrack2017 test | mAP: 74.9 -> 77.9 | Helped |
| Use PoseWarper instead of the strongest listed baseline | Multi-person pose estimation · PoseTrack2018 val | mAP: 75.0 -> 79.7 | Helped |
| Use PoseWarper instead of the strongest listed baseline | Multi-person pose estimation · PoseTrack2018 test | mAP: 76.4 -> 78.0 | Helped |
| Add dilated convolution blocks (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 87.2 -> 88.7 | Helped |
| Add a second dilation level, using rates 3,6 instead of only rate 3 (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 87.0 -> 88.0 | Helped |
| Add a third dilation level, using rates 3,6,12 (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 88.0 -> 88.4 | Helped |
| Add a fourth dilation level, using rates 3,6,12,18 (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 88.4 -> 88.6 | Helped |
| Add a fifth dilation level, using rates 3,6,12,18,24 (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 88.6 -> 88.7 | Helped |
| Add more than five dilated convolutions (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP | No clear effect |
| Replace dilation rates 3,6,12,18,24 with 1,2,3,4,5 (ablation) | PoseWarper · PoseTrack2017 val, video pose propagation | mAP: 88.7 -> 88.6 | Hurt |
Closest papers by meaning
- Deep Dual Consecutive Network for Human Pose Estimation (CVPR 2021 · similarity 0.91)
- Temporal Feature Alignment and Mutual Information Maximization for Video-Based Human Pose Estimation (CVPR 2022 · similarity 0.91)
- End-to-End Multi-Person Pose Estimation with Pose-Aware Video Transformer (AAAI 2026 · similarity 0.90)
- SpatioTemporal Learning for Human Pose Estimation in Sparsely-Labeled Videos (AAAI 2025 · similarity 0.88)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.88)
- Can Generative Video Models Help Pose Estimation? (CVPR 2025 · 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.