Browse › Computer Vision › 3D Human Pose Estimation › MPI-INF-3DHP
MPI-INF-3DHP Benchmark (3D Human Pose Estimation)
3D Human Pose Estimation is a computer vision task that involves estimating the 3D positions and orientations of body joints and bones from 2D images or videos. The goal is to reconstruct the 3D pose of a person in real-time, which can be used in a variety of applications, such as virtual reality, human-computer interaction, and motion analysis.
The archive carries no text for this table; the description above is the archive's text for the task 3D Human Pose Estimation. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Direction inferred from the metric name, not from the archive: AUC (higher is better), Acceleration Error (lower is better). Not inferred (points only, no best-so-far line): MPJPE, PCK, 3DPCK, PA-MPJPE. Points are placed at the row's paper date; 108 of 108 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | TCPFormer (T=81) | 15 | 87.7 | 99.0 | – | Paper | Code | 2025 | linked, not harvested | report | |||
| 2 | MotionAGFormer-L (T=81) | 16.2 | 85.3 | 98.2 | – | Paper | Code | 2023 | 14 of 16 ran · 2 unverified | report | |||
| 3 | KTPFormer | 16.7 | 85.9 | 98.9 | – | Paper | Code | 2024 | 6 of 10 ran · 4 unverified | report | |||
| 4 | MotionAGFormer-S (T=81) | 17.1 | 84.5 | 98.3 | – | Paper | Code | 2023 | 14 of 16 ran · 2 unverified | report | |||
| 5 | TCPFormer (T=27) | 17.8 | 86.5 | 98.7 | – | Paper | Code | 2025 | linked, not harvested | report | |||
| 6 | MotionAGFormer-B (T=81) | 18.2 | 84.2 | 98.3 | – | Paper | Code | 2023 | 14 of 16 ran · 2 unverified | report | |||
| 7 | MotionAGFormer-XS (T=27) | 19.2 | 83.5 | 98.2 | – | Paper | Code | 2023 | 14 of 16 ran · 2 unverified | report | |||
| 8 | STCFormer (T=81) | 23.1 | 83.9 | 98.7 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 9 | GLA-GCN (T=81) | 27.76 | 79.12 | 98.53 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 10 | PoseFormerV2 (T=81) | 27.8 | 78.8 | 97.9 | – | Paper | Code | 2023 | 5 of 6 ran · 1 unverified | report | |||
| 11 | Fusionformer (f=9) | 28.2 | 70 | 97.9 | – | Paper | – | 2022 | no code linked | report | |||
| 12 | HSTFormer (T=81) | 28.3 | 78.6 | 98 | – | Paper | – | 2023 | no code linked | report | |||
| 13 | DiffPose | 29.1 | 75.9 | 98 | – | Paper | Code | 2022 | 7 of 14 ran · 7 unverified | report | |||
| 14 | D3DP (N=243, H=20, K=20, J-Agg) | 29.7 | 78.2 | 97.7 | – | Paper | Code | 2023 | 0 of 2 ran · 2 unverified | report | |||
| 15 | P-STMO (N=81) | 32.2 | 75.8 | 97.9 | – | Paper | Code | 2022 | 9 of 11 ran · 2 unverified | report | |||
| 16 | LMT R152 384x384 | 33.7 | 77.09 | 99.37 | – | Paper | – | 2022 | no code linked | report | |||
| 17 | RTPCA | 40.5 | 74.2 | 98.8 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 18 | U-CondDGConv | 42.5 | 69.5 | 97.9 | – | Paper | Code | 2021 | 2 of 3 ran · 1 unverified | report | |||
| 19 | (R50-224) LMT | 45.87 | 71.57 | 96.59 | – | Paper | – | 2022 | no code linked | report | |||
| 20 | Ray3D (T=9 CPN H36M+HEva+3DHP) | 46.6 | – | Paper | Code | 2022 | 4 of 5 ran · 1 unverified | report | |||||
| 21 | ConvFormer | 53.6 | 69.8 | 96.4 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 22 | MixSTE (T=27) | 54.9 | 66.5 | 94.4 | – | Paper | Code | 2022 | 4 of 4 ran · 0 unverified | report | |||
| 23 | ZeDO (S=50) | 55.2 | 65.6 | 93 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 24 | MixSTE (T=1) | 57.9 | 63.8 | 94.2 | – | Paper | Code | 2022 | 4 of 4 ran · 0 unverified | report | |||
| 25 | MHFormer | 58 | 63.3 | 93.8 | – | Paper | Code | 2021 | 4 of 10 ran · 6 unverified | report | |||
| 26 | PoseDA | 61.3 | 62.5 | 92.1 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 27 | PLIKS (HR48) | 67.6 | 54.1 | 93.9 | – | Paper | Code | 2022 | linked, not harvested | report | |||
| 28 | UGCN | 68.1 | 62.1 | 86.9 | – | Paper | Code | 2020 | 2 of 9 ran · 7 unverified | report | |||
| 29 | PoseAug (+Extra2D) | 71.1 | 57.9 | 89.2 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 30 | ARTS (Resnet50 L=16) | 71.8 | 53 | 7.4 | – | Paper | Code | 2024 | linked, not harvested | report | |||
| 31 | VPose+PoseAug | 73 | 57.3 | 88.6 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 32 | HR-Net+VPose+PoseAug | 73.2 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 33 | DG-Net (T=4) | 76 | 53.8 | – | Paper | – | 2021 | no code linked | report | ||||
| 34 | CrossFormer | 76.3 | 57.5 | 89.1 | – | Paper | Code | 2022 | linked, not harvested | report | |||
| 35 | HR-Net+ST-GCN+PoseAug | 76.6 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 36 | PoseFormer (9 frames) | 77.1 | 56.4 | 88.6 | – | Paper | Code | 2021 | 4 of 8 ran · 4 unverified | report | |||
| 37 | Anatomy3D (T=81) | 78.8 | 54 | 87.9 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 38 | PoseGU | 79.1 | 55.1 | 86.3 | – | Paper | – | 2022 | no code linked | report | |||
| 39 | Anatomy3D (T=243) | 79.1 | 53.8 | 87.8 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 40 | PoseTriplet | 79.5 | 53.1 | 89.1 | – | Paper | Code | 2022 | linked, not harvested | report | |||
| 41 | Trajectory Space Factorization (F=25) | 79.8 | 51.4 | 83.6 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 42 | MAED | 83.6 | 56.2 | – | Paper | Code | 2021 | 0 of 18 ran · 18 unverified | report | ||||
| 43 | TAR (N=9) | 85.9 | 60.5 | 9.2 | – | Paper | – | 2023 | no code linked | report | |||
| 44 | HybrIK-Transformer (HrNet-48) | 86.2 | 48.9 | 88.6 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 45 | IKOL | 88.8 | 48.1 | 87.9 | – | Paper | Code | 2023 | 0 of 7 ran · 7 unverified | report | |||
| 46 | HybrIK (HRNet-W48) | 91 | 47.3 | 87.1 | – | Paper | Code | 2023 | 0 of 8 ran · 8 unverified | report | |||
| 47 | HybrIK | 91.0 | 46.9 | 87.5 | – | Paper | Code | 2020 | 1 of 1 ran · 0 unverified | report | |||
| 48 | MargiPose (multi-crop) | 91.3 | 47 | 85.4 | – | Paper | Code | 2018 | linked, not harvested | report | |||
| 49 | RepNet (H36M) | 92.5 | 54.8 | 81.8 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 50 | Multi-view Temporal self-supervised | 93.0 | 50.1 | 81.0 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 51 | DST-VIBE | 93.4 | 62.2 | 11.9 | – | Paper | – | 2021 | no code linked | report | |||
| 52 | GLoT | 93.9 | 61.5 | 7.9 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 53 | TAPE (T=16) | 94 | 56.7 | 12.4 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 54 | Self-Attentive | 94.3 | 90.1 | 60.7 | – | Paper | – | 2021 | no code linked | report | |||
| 55 | LMR | 94.6 | 62.4 | – | Paper | – | 2021 | no code linked | report | ||||
| 56 | STR | 95.3 | 61.6 | 8.4 | – | Paper | – | 2022 | no code linked | report | |||
| 57 | TePose (T=6 3DPW) | 96.2 | 63.1 | 16.7 | – | Paper | Code | 2022 | linked, not harvested | report | |||
| 58 | VIBE | 96.6 | 89.3 | 64.6 | – | Paper | Code | 2019 | 0 of 1 ran · 1 unverified | report | |||
| 59 | MPS-Net (T=16) | 96.7 | 62.8 | 9.6 | – | Paper | – | 2022 | no code linked | report | |||
| 60 | SSP-Net | 96.8 | 44.3 | 83.2 | – | Paper | – | 2020 | no code linked | report | |||
| 61 | DC-GNet | 97.2 | 40.7 | 62.5 | – | Paper | – | 2021 | no code linked | report | |||
| 62 | TCMR (T=16 w/o H3.6M) | 97.3 | 63.5 | 8.5 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 63 | TCMR (T=16 w/o 3DPW) | 97.4 | 62.8 | 8 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 64 | MoCap-SPIN + PoseBERT | 97.4 | 63.3 | 8.7 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 65 | RepNet (3DHP) | 97.8 | 58.5 | 82.5 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 66 | DDT | 97.8 | 65.4 | 8.2 | – | Paper | – | 2023 | no code linked | report | |||
| 67 | XNect (SelecSLS) | 98.4 | 45.3 | 82.8 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 68 | EvoSkeleton | 99.7 | 46.1 | 81.2 | – | Paper | Code | 2020 | 2 of 2 ran · 0 unverified | report | |||
| 69 | DynaBOA | 101.5 | 43.1 | 79.5 | 66.1 | – | Paper | Code | 2021 | linked, not harvested | report | ||
| 70 | RSC-Net | 103.36 | 70.01 | – | Paper | Code | 2020 | linked, not harvested | report | ||||
| 71 | CanonPose | 104 | 77 | – | Paper | Code | 2020 | linked, not harvested | report | ||||
| 72 | DSR | 104.7 | 66.7 | – | Paper | Code | 2021 | 3 of 3 ran · 0 unverified | report | ||||
| 73 | SPIN | 105.2 | 37.1 | 76.4 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 74 | EpipolarPose (fully-supervised) | 108.99 | 77.5 | – | Paper | Code | 2019 | linked, not harvested | report | ||||
| 75 | XFormer (HRNet) | 109.8 | 64.5 | – | Paper | – | 2023 | no code linked | report | ||||
| 76 | GeoRep (fully-supervised) | 110.8 | 80.2 | – | Paper | – | 2020 | no code linked | report | ||||
| 77 | Pose Consensus (monocular) | 112.1 | 42.1 | 80.6 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 78 | PONet | 115.0 | 40.6 | 76.1 | – | Paper | – | 2021 | no code linked | report | |||
| 79 | Mehta | 117.6 | 39.3 | 75.7 | – | Paper | – | 2016 | no code linked | report | |||
| 80 | Single-Shot Multi-Person | 122.2 | 37.8 | 75.2 | – | Paper | Code | 2017 | 1 of 2 ran · 1 unverified | report | |||
| 81 | HMR | 124.2 | 36.5 | 72.9 | 89.8 | – | Paper | Code | 2017 | linked, not harvested | report | ||
| 82 | VNect (Augm.) | 124.7 | 40.4 | 76.6 | – | Paper | Code | 2017 | linked, not harvested | report | |||
| 83 | Explicit Compositional Depth Maps | 62.4 | 93.2 | – | Paper | – | 2020 | no code linked | report | ||||
| 84 | DC-GCT | 55.9 | 87.5 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 85 | SPIN (Rigid Alignment) | 55.6 | 92.5 | 67.5 | – | Paper | Code | 2019 | linked, not harvested | report | |||
| 86 | HTNet | 54.1 | 86.7 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 87 | Modulated-GCN | 53.7 | 86.1 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 88 | Regular Splitting Graph Network | 53.2 | 85.6 | – | Paper | Code | 2023 | linked, not harvested | report | ||||
| 89 | HGN | 52.1 | 85.2 | – | Paper | Code | 2021 | linked, not harvested | report | ||||
| 90 | TIPSy-GAN (GT) | 48.8 | 78 | – | Paper | – | 2022 | no code linked | report | ||||
| 91 | Skeletal GNN | 46.2 | 82.1 | – | Paper | – | 2021 | no code linked | report | ||||
| 92 | Graph Stacked Hourglass Network | 45.8 | 80.1 | – | Paper | Code | 2021 | 12 of 15 ran · 3 unverified | report | ||||
| 93 | PoseLifter | 45.1 | 84 | – | Paper | Code | 2019 | linked, not harvested | report | ||||
| 94 | SRNET | 43.8 | 77.6 | – | Paper | Code | 2020 | 1 of 9 ran · 8 unverified | report | ||||
| 95 | ContextPose | 42.7 | 80.5 | – | Paper | Code | 2021 | 1 of 1 ran · 0 unverified | report | ||||
| 96 | VNect (ResNet 50 GT) | 41.6 | 79.4 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 97 | Mehta | 40.8 | 64.7 | – | Paper | – | 2016 | no code linked | report | ||||
| 98 | HEMlets Pose | 38 | 75.3 | – | Paper | – | 2019 | no code linked | report | ||||
| 99 | 2D-3D Lifting Network | 36.3 | 71.1 | – | Paper | – | 2019 | no code linked | report | ||||
| 100 | Ordinal Depth Supervision | 35.3 | 71.9 | – | Paper | Code | 2018 | linked, not harvested | report | ||||
| 101 | Stereoscopic View Synthesis Subnetwork | 33.8 | 71.2 | – | Paper | Code | 2019 | linked, not harvested | report | ||||
| 102 | OriNet | 32.1 | 64.6 | – | Paper | Code | 2018 | linked, not harvested | report | ||||
| 103 | Adversarial Learning | 32.0 | 69.0 | – | Paper | – | 2018 | no code linked | report | ||||
| 104 | Probabilistic Monocular | 84.3 | – | Paper | Code | 2021 | 0 of 9 ran · 9 unverified | report | |||||
| 105 | Spatio-Temporal Network (T=128) | 84.1 | – | Paper | – | 2020 | no code linked | report | |||||
| 106 | WSGAN | 79.3 | – | Paper | Code | 2020 | 0 of 11 ran · 11 unverified | report | |||||
| 107 | MDM | 67.9 | – | Paper | Code | 2019 | 1 of 11 ran · 10 unverified | report | |||||
| 108 | EFT | 67.5 | – | Paper | Code | 2020 | 0 of 1 ran · 1 unverified | report |
All 108 rows shown. 108 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 31 rows have a graph line, from 27 distinct papers; 23 rows (19 papers) have at least one sample that ran. Counting each paper once: Syntology ran 83 of 197 samples; 114 unverified. Separately, 44 of those 197 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections