Research map / 3D vision / Human pose estimation
DeeperCut: A Deeper, Stronger, and Faster Multi-person Pose Estimation Model
Eldar Insafutdinov, Leonid Pishchulin, Bjoern Andres, Mykhaylo Andriluka, Bernt Schiele
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 intermediate supervision (ablation) | ResNet-152 part detector · LSP (PC) | PCK: 89.1 -> 90.1 | Improved |
| add intermediate supervision (ablation) | ResNet-152 part detector · MPII Single Person | PCKh: 87.8 -> 88.5 | Improved |
| use ResNet-152 instead of ResNet-50 (16 px stride) (ablation) | part detector · LSP (PC) | PCK: 87.2 -> 89.1 | Improved |
| increase stride from 8 px to 16 px with 2x up-sampling (ablation) | ResNet-50 · LSP (PC) | PCK: 87.8 -> 87.2 | Hurt |
| replace DeepCut pairwise with image-conditioned pairwise (ablation) | DeepCut · MPII Multi-Person Val | AP: 33.3 -> 47.7 | Improved |
| replace DeepCut unary with proposed ResNet part detectors (ablation) | DeeperCut · MPII Multi-Person Val | AP: 47.7 -> 52.3 | Improved |
| apply location refinement before NMS (ablation) | DeeperCut · MPII Multi-Person Val | AP: 52.3 -> 52.6 | Improved |
| make pairwise bi-directional instead of uni-directional (ablation) | image-conditioned pairwise · MPII Multi-Person Val | AP: 51.1 -> 52.6 | Improved |
| add angle features to pairwise (ablation) | image-conditioned pairwise · MPII Multi-Person Val | AP: 51.3 -> 52.6 | Improved |
| use more aggressive NMS (24 px radius) (ablation) | 1-stage optimization · MPII Multi-Person Val | AP: 52.6 -> 54.5 | Improved |
| increase detections from 100 to 150 (ablation) | 1-stage optimization · MPII Multi-Person Val | AP: 54.5 -> 55.1 | Improved |
| use 2-stage incremental optimization instead of 1-stage with 150 detections (ablation) | DeeperCut · MPII Multi-Person Val | AP: 55.1 -> 56.5 | Improved |
Later papers that compared with this method
5 later accepted papers report DeeperCut as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019 · better on 2 of 2 shared results)
- Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (CVPR 2017 · better on 3 of 3 shared results)
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · better on 2 of 2 shared results)
- PoseTrack: Joint Multi-Person Pose Estimation and Tracking (CVPR 2017 · better on 1 of 1 shared result)
- RMPE: Regional Multi-Person Pose Estimation (ICCV 2017 · better on 1 of 1 shared result)
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
- DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation (CVPR 2016 · similarity 0.95)
- Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation (AAAI 2020 · similarity 0.93)
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · similarity 0.91)
- Group Pose: A Simple Baseline for End-to-End Multi-Person Pose Estimation (ICCV 2023 · similarity 0.91)
- Dual Networks Based 3D Multi-Person Pose Estimation From Monocular Video (TPAMI 2022 · similarity 0.91)
- Monocular 3D Pose and Shape Estimation of Multiple People in Natural Scenes - The Importance of Multiple Scene Constraints (CVPR 2018 · similarity 0.90)
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.
