Multi-person Pose Estimation on COCO 2017 (test-dev)
78.1APMSPN
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| MSPNBackbone=4xRes-50, Input Size=384×288, Ensemble=true, External Data=true2019.01 | 78.1 | 94.1 | 85.9 | 74.5 | 83.3 | 83.1 | 96.7 | 89.8 | 79.3 | 88.2 | |
| MSPNBackbone=4xRes-50, Input Size=384×288, External Data=true, Ensemble=false2019.01 | 77.1 | 93.8 | 84.6 | 73.4 | 82.3 | 82.3 | 96.5 | 88.9 | 78.4 | 87.7 | |
| HRNetBackbone=HRNet-W48, Input Size=384×288, External Data=true2019.01 | 77 | 92.7 | 84.5 | 73.4 | 83.1 | 82 | — | — | — | — | |
| Simple BaseBackbone=Res-152, Input Size=384×288, Ensemble=true, External Data=true2019.01 | 76.5 | 92.4 | 84 | 73 | 82.7 | 81.5 | 95.8 | 88.2 | 77.4 | 87.2 | |
| MSPNBackbone=4xRes-50, Input Size=384×288, Ensemble=false, External Data=false2019.01 | 76.1 | 93.4 | 83.8 | 72.3 | 81.5 | 81.6 | 96.3 | 88.1 | 77.5 | 87.1 | |
| HRNetBackbone=HRNet-W48, Input Size=384×2882019.01 | 75.5 | 92.5 | 83.3 | 71.9 | 81.5 | 80.5 | — | — | — | — | |
| HRNet-W48category=Top-down2020.07 | 75.5 | 92.5 | 83.3 | 71.9 | 81.5 | 80.5 | — | — | — | — | |
| GroupPoseBackbone=Swin-L, #Params=219.6M, Time (ms)=214, End-to-end=true2025.06 | 74.8 | 91.6 | 82.1 | 69.4 | 83 | — | — | — | — | — | |
| Simple BaseBackbone=Res-152, Input Size=384×2882019.01 | 73.7 | 91.9 | 81.1 | 70.3 | 80 | 79 | — | — | — | — | |
| SBLcategory=Top-down2020.07 | 73.7 | 91.9 | 81.1 | 70.3 | 80 | 79 | — | — | — | — | |
| QueryPoseBackbone=Swin-L, End-to-end=true2025.06 | 73.3 | 91.3 | 79.5 | 68.5 | 81.2 | — | — | — | — | — | |
| RTMO-lBackbone=CSPDarknet, #Params=44.8M, Time (ms)=19.1, additional_training_data=true2023.12 | 73.3 | 91.9 | — | 68.3 | 81.1 | 77.4 | — | — | — | — | |
| CPNBackbone=Res-Inception, Input Size=384×288, Ensemble=true2019.01 | 73 | 91.7 | 80.9 | 69.5 | 78.1 | 79 | 95.1 | 85.9 | 74.8 | 84.6 | |
| ED-PoseBackbone=Swin-L, #Params=218.0M, Time (ms)=222.6, End-to-end=true2025.06 | 72.7 | 92.3 | 80.9 | 67.6 | 80 | — | — | — | — | — | |
| ED-PoseBackbone=Swin-L, #Params=218.0M, Time (ms)=265.62023.12 | 72.7 | 92.3 | 80.9 | 67.6 | 80 | — | — | — | — | — | |
| CFNcategory=Top-down2020.07 | 72.6 | 86.1 | 69.7 | 78.3 | 64.1 | — | — | — | — | — | |
| RMPEcategory=Top-down2020.07 | 72.3 | 89.2 | 79.1 | 68 | 78.6 | — | — | — | — | — | |
| DETRPose-XBackbone=HGNetv2-B5, #Params=82.3M, Time (ms)=52.1, End-to-end=true2025.06 | 72.2 | 91.4 | 79.3 | 67 | 80.6 | 78.8 | — | — | — | — | |
| CPNBackbone=Res-Inception, Input Size=384×2882019.01 | 72.1 | 91.4 | 80 | 68.7 | 77.2 | 78.5 | 95.1 | 85.3 | 74.2 | 84.3 | |
| CPNcategory=Top-down2020.07 | 72.1 | 91.4 | 80 | 68.7 | 77.2 | 78.5 | — | — | — | — | |
| RTMO-LBackbone=CSPDarknet, #Params=44.8M, Time (ms)=19.1, End-to-end=false2025.06 | 71.6 | 91.1 | 79 | 66.8 | 79.1 | 75.6 | — | — | — | — | |
| RTMO-lBackbone=CSPDarknet, #Params=44.8M, Time (ms)=19.12023.12 | 71.6 | 91.1 | 79 | 66.8 | 79.1 | 75.6 | — | — | — | — | |
| RTMO-mBackbone=CSPDarknet, #Params=22.6M, Time (ms)=12.4, additional_training_data=true2023.12 | 71.5 | 91 | 78.6 | 66.1 | 79.9 | 75.6 | — | — | — | — | |
| DETRPose-LBackbone=HGNetv2-B4, #Params=36.8M, Time (ms)=32.5, End-to-end=true2025.06 | 71.2 | 91.2 | 78.1 | 65.8 | 79.9 | 78.1 | — | — | — | — | |
| RTMO-r50Backbone=ResNet-50, #Params=41.7M, Time (ms)=15.5, End-to-end=false2025.06 | 70.9 | 91 | 78.2 | 65.8 | 79.1 | 75 | — | — | — | — | |
| RTMO-r50Backbone=ResNet-50, #Params=41.7M, Time (ms)=15.52023.12 | 70.9 | 91 | 78.2 | 65.8 | 79.1 | 75 | — | — | — | — | |
| CIDBackbone=HRNet-w48, #Params=65.4M, Time (ms)=94.8, End-to-end=false2025.06 | 70.7 | 90.4 | 77.9 | 66.3 | 77.8 | 76.4 | — | — | — | — | |
| CIDBackbone=HRNet-w48, #Params=65.4M, Time (ms)=94.82023.12 | 70.7 | 90.4 | 77.9 | 66.3 | 77.8 | 76.4 | — | — | — | — | |
| PETRBackbone=Swin-L, #Params=213.8M, Time (ms)=133, End-to-end=true2025.06 | 70.5 | 91.5 | 78.7 | 65.2 | 78 | — | — | — | — | — | |
| PETRBackbone=Swin-L, #Params=213.8M, Time (ms)=1332023.12 | 70.5 | 91.5 | 78.7 | 65.2 | 78 | — | — | — | — | — | |
| KAPAO-LBackbone=CSPNet, #Params=77.0M, Time (ms)=50.2, End-to-end=false2025.06 | 70.3 | 91.2 | 77.8 | 66.3 | 76.8 | 77.7 | — | — | — | — | |
| KAPAO-lBackbone=CSPNet, #Params=77.0M, Time (ms)=50.22023.12 | 70.3 | 91.2 | 77.8 | 66.3 | 76.8 | 77.7 | — | — | — | — | |
| YOLO-Pose-lBackbone=CSPDarknet, #Params=61.3M, Time (ms)=20.5, End-to-end=false2025.06 | 70.2 | 91.1 | 77.8 | 65.3 | 78.2 | 74.3 | — | — | — | — | |
| GroupPoseBackbone=ResNet-50, #Params=52.4M, Time (ms)=116.5, End-to-end=true2025.06 | 70.2 | 90.5 | 77.8 | 64.7 | 78 | — | — | — | — | — | |
| YOLO-Pose-lBackbone=CSPDarknet, #Params=61.3M, Time (ms)=20.52023.12 | 70.2 | 91.1 | 77.8 | 65.3 | 78.2 | 74.3 | — | — | — | — | |
| RTMO-MBackbone=CSPDarknet, #Params=22.6M, Time (ms)=12.4, End-to-end=false2025.06 | 70.1 | 90.6 | 77.1 | 65.1 | 78.1 | 74.2 | — | — | — | — | |
| RTMO-mBackbone=CSPDarknet, #Params=22.6M, Time (ms)=12.42023.12 | 70.1 | 90.6 | 77.1 | 65.1 | 78.1 | 74.2 | — | — | — | — | |
| ED-PoseBackbone=ResNet-50, #Params=50.6M, Time (ms)=125.5, End-to-end=true2025.06 | 69.8 | 90.2 | 77.2 | 64.3 | 77.4 | — | — | — | — | — | |
| ED-PoseBackbone=ResNet-50, #Params=50.6M, Time (ms)=135.22023.12 | 69.8 | 90.2 | 77.2 | 64.3 | 77.4 | — | — | — | — | — | |
| CIDBackbone=HRNet-w32, #Params=29.4M, Time (ms)=84, End-to-end=false2025.06 | 68.9 | 89.9 | 76.9 | 63.2 | 77.7 | 74.6 | — | — | — | — | |
| CIDBackbone=HRNet-w32, #Params=29.4M, Time (ms)=84.02023.12 | 68.9 | 89.9 | 76.9 | 63.2 | 77.7 | 74.6 | — | — | — | — | |
| KAPAO-MBackbone=CSPNet, #Params=35.8M, Time (ms)=37, End-to-end=false2025.06 | 68.8 | 90.5 | 76.5 | 64.3 | 76 | 76.3 | — | — | — | — | |
| KAPAO-mBackbone=CSPNet, #Params=35.8M, Time (ms)=37.02023.12 | 68.8 | 90.5 | 76.5 | 64.3 | 76 | 76.3 | — | — | — | — | |
| QueryPoseBackbone=ResNet-50, End-to-end=true2025.06 | 68.7 | 88.6 | 74.4 | 63.8 | 76.5 | — | — | — | — | — | |
| PersonLabcategory=Bottom-up, multi-scale test=true2020.07 | 68.7 | 89 | 75.4 | 64.1 | 75.5 | 75.4 | — | — | — | — | |
| YOLO-Pose-mBackbone=CSPDarknet, #Params=29.3M, Time (ms)=12.5, End-to-end=false2025.06 | 68.6 | 90.7 | 75.8 | 63.4 | 77.1 | 72.8 | — | — | — | — | |
| YOLO-Pose-mBackbone=CSPDarknet, #Params=29.3M, Time (ms)=12.52023.12 | 68.6 | 90.7 | 75.8 | 63.4 | 77.1 | 72.8 | — | — | — | — | |
| DETRPose-MBackbone=HGNetv2-B2, #Params=23.5M, Time (ms)=25.8, End-to-end=true2025.06 | 68.5 | 88.6 | 74.6 | 62.1 | 78.3 | 74.6 | — | — | — | — | |
| HigherHRNet-W48Input Size=640x640, #Params=63.8M, #MACs=155.1G, Nano Latency (ms)=2101, Mobile Latency (ms)=1532, Pi Latency (ms)=123022022.05 | 68.4 | — | — | — | — | — | — | — | — | — | |
| IPRcategory=Top-down2020.07 | 67.8 | 88.2 | 74.8 | 63.9 | 74 | — | — | — | — | — | |
| RTMO-sBackbone=CSPDarknet, #Params=9.9M, Time (ms)=8.9, additional_training_data=true2023.12 | 67.7 | 89.4 | 74.5 | 61.5 | 77.2 | 71.9 | — | — | — | — | |
| PETRBackbone=ResNet-50, #Params=43.7M, Time (ms)=89, End-to-end=true2025.06 | 67.6 | 89.8 | 75.3 | 61.6 | 76 | — | — | — | — | — | |
| HGGcategory=Bottom-up, multi-scale test=true, single-person pose refinement=false2020.07 | 67.6 | 85.1 | 73.7 | 62.7 | 74.6 | 71.3 | — | — | — | — | |
| PETRBackbone=ResNet-50, #Params=43.7M, Time (ms)=892023.12 | 67.6 | 89.8 | 75.3 | 61.6 | 76 | — | — | — | — | — | |
| RTMO-SBackbone=CSPDarknet, #Params=9.9M, Time (ms)=8.9, End-to-end=false2025.06 | 66.9 | 88.8 | 73.6 | 61.1 | 75.7 | 70.9 | — | — | — | — | |
| SPMcategory=Bottom-up, single-person pose refinement=true, multi-scale test=true2020.07 | 66.9 | 88.5 | 72.9 | 62.6 | 73.1 | — | — | — | — | — | |
| RTMO-sBackbone=CSPDarknet, #Params=9.9M, Time (ms)=8.92023.12 | 66.9 | 88.8 | 73.6 | 61.1 | 75.7 | 70.9 | — | — | — | — | |
| DETRPose-SBackbone=HGNetv2-B0, #Params=11.9M, Time (ms)=20.3, End-to-end=true2025.06 | 66.5 | 87.4 | 72.5 | 59.8 | 76.8 | 73 | — | — | — | — | |
| PersonLabInput Size=1401x1401, #Params=68.7M, #MACs=405.5G2022.05 | 66.5 | — | — | — | — | — | — | — | — | — | |
| InsPoseBackbone=ResNet-101, Time (ms)=100, End-to-end=false2025.06 | 66.3 | 89.2 | 73 | 61.2 | 73.9 | — | — | — | — | — | |
| InsPoseBackbone=ResNet-101, Time (ms)=1002023.12 | 66.3 | 89.2 | 73 | 61.2 | 73.9 | — | — | — | — | — | |
| FCPoseBackbone=ResNet-101, #Params=60.5M, Time (ms)=93, End-to-end=false2025.06 | 65.6 | 87.9 | 72.6 | 62.1 | 72.3 | — | — | — | — | — | |
| FCPoseBackbone=ResNet-101, #Params=60.5M, Time (ms)=932023.12 | 65.6 | 87.9 | 72.6 | 62.1 | 72.3 | — | — | — | — | — | |
| Associative Embeddingtraining data=excluding external data2016.11 | 65.5 | 86.8 | 72.3 | 60.6 | 72.6 | 70.2 | 89.5 | 76 | 64.6 | 78.1 | |
| AEBackbone=Not specified, Input Size=512×5122019.01 | 65.5 | 86.8 | 72.3 | 60.6 | 72.6 | 70.2 | 89.5 | 76 | 64.6 | 78.1 | |
| AEcategory=Bottom-up, single-person pose refinement=true, multi-scale test=true2020.07 | 65.5 | 86.8 | 72.3 | 60.6 | 72.6 | 70.2 | — | — | — | — | |
| InsPoseBackbone=ResNet-50, #Params=50.2M, Time (ms)=80, End-to-end=false2025.06 | 65.4 | 88.9 | 71.7 | 60.2 | 72.7 | — | — | — | — | — | |
| InsPoseBackbone=ResNet-50, #Params=50.2M, Time (ms)=802023.12 | 65.4 | 88.9 | 71.7 | 60.2 | 72.7 | — | — | — | — | — | |
| G-RMItraining data=excluding external data2016.11 | 64.9 | 85.5 | 71.3 | 62.3 | 70 | 69.7 | 88.7 | 75.5 | 64.4 | 77.1 | |
| G-RMIBackbone=Res-152, Input Size=353×2572019.01 | 64.9 | 85.5 | 71.3 | 62.3 | 70 | 69.7 | 88.7 | 75.5 | 64.4 | 77.1 | |
| G-RMIcategory=Top-down2020.07 | 64.9 | 85.5 | 71.3 | 62.3 | 70 | 69.7 | — | — | — | — | |
| G-RMI PoseInitialization=JFT-300M2017.07 | 64.8 | 85.8 | — | — | — | 69.4 | 88.4 | — | — | — | |
| Directposecategory=Bottom-up, multi-scale test=true2020.07 | 64.8 | 87.8 | 71.1 | 60.4 | 71.5 | — | — | — | — | — | |
| G-RMI PoseInitialization=ImageNet + JFT-300M2017.07 | 64.4 | 85.7 | — | — | — | 69.1 | 88.2 | — | — | — | |
| FCPoseBackbone=ResNet-50, #Params=41.7M, Time (ms)=68, End-to-end=false2025.06 | 64.3 | 87.3 | 71 | 61.6 | 70.5 | — | — | — | — | — | |
| FCPoseBackbone=ResNet-50, #Params=41.7M, Time (ms)=682023.12 | 64.3 | 87.3 | 71 | 61.6 | 70.5 | — | — | — | — | — | |
| KAPAO-SBackbone=CSPNet, #Params=12.6M, Time (ms)=26.9, End-to-end=false2025.06 | 63.8 | 88.4 | 70.4 | 58.6 | 71.7 | 71.2 | — | — | — | — | |
| KAPAO-sBackbone=CSPNet, #Params=12.6M, Time (ms)=26.92023.12 | 63.8 | 88.4 | 70.4 | 58.6 | 71.7 | 71.2 | — | — | — | — | |
| DirectPoseBackbone=ResNet-101, End-to-end=false2025.06 | 63.3 | 86.7 | 69.4 | 57.8 | 71.2 | — | — | — | — | — | |
| DirectPoseBackbone=ResNet-1012023.12 | 63.3 | 86.7 | 69.4 | 57.8 | 71.2 | — | — | — | — | — | |
| YOLO-Pose-sBackbone=CSPDarknet, #Params=10.8M, Time (ms)=7.9, End-to-end=false2025.06 | 63.2 | 87.8 | 69.5 | 57.6 | 72.6 | 67.6 | — | — | — | — | |
| YOLO-Pose-sBackbone=CSPDarknet, #Params=10.8M, Time (ms)=7.92023.12 | 63.2 | 87.8 | 69.5 | 57.6 | 72.6 | 67.6 | — | — | — | — | |
| Mask R-CNNBackbone=Res-50-FPN, Input Size=Not specified2019.01 | 63.1 | 87.3 | 68.7 | 57.8 | 71.4 | — | — | — | — | — | |
| Mask-RCNNcategory=Top-down2020.07 | 63.1 | 87.3 | 68.7 | 57.8 | 71.4 | 69.7 | — | — | — | — | |
| CenterNetBackbone=Hourglass, #Params=194.9M, Time (ms)=160, End-to-end=false2025.06 | 63 | 86.8 | 69.6 | 58.9 | 70.4 | — | — | — | — | — | |
| CenterNetBackbone=Hourglass, #Params=194.9M, Time (ms)=1602023.12 | 63 | 86.8 | 69.6 | 58.9 | 70.4 | — | — | — | — | — | |
| Mask-RCNNtraining data=excluding external data2016.11 | 62.7 | 87 | 68.4 | 57.4 | 71.1 | — | — | — | — | — | |
| G-RMI PoseInitialization=ImageNet2017.07 | 62.4 | 84 | — | — | — | 66.7 | 86.6 | — | — | — | |
| DirectPoseBackbone=ResNet-50, Time (ms)=74, End-to-end=false2025.06 | 62.2 | 86.4 | 68.2 | 56.7 | 69.8 | — | — | — | — | — | |
| DirectPoseBackbone=ResNet-50, Time (ms)=742023.12 | 62.2 | 86.4 | 68.2 | 56.7 | 69.8 | — | — | — | — | — | |
| CMU-Posetraining data=excluding external data2016.11 | 61.8 | 84.9 | 67.5 | 57.1 | 68.2 | 66.5 | 87.2 | 71.8 | 60.6 | 74.6 | |
| CMU Pose2017.07 | 61.8 | 84.9 | — | — | — | 66.5 | 87.2 | — | — | — | |
| CMU PoseBackbone=Not specified, Input Size=Not specified2019.01 | 61.8 | 84.9 | 67.5 | 57.1 | 68.2 | 66.5 | 87.2 | 71.8 | 60.6 | 74.6 | |
| OpenPosecategory=Bottom-up, single-person pose refinement=true2020.07 | 61.8 | 84.9 | 67.5 | 57.1 | 68.2 | 66.5 | — | — | — | — | |
| LitePose-SInput Size=448x448, #Params=2.7M, #MACs=5.0G, Nano Latency (ms)=97, Mobile Latency (ms)=76, Pi Latency (ms)=4202022.05 | 56.7 | — | — | — | — | — | — | — | — | — | |
| HourglassInput Size=512x512, #Params=277.8M, #MACs=206.9G2022.05 | 56.6 | — | — | — | — | — | — | — | — | — | |
| EfficientHRNet-H-2Input Size=448x448, #Params=8.3M, #MACs=7.9G, Nano Latency (ms)=170, Mobile Latency (ms)=182, Pi Latency (ms)=8782022.05 | 52.8 | — | — | — | — | — | — | — | — | — | |
| LitePose-XSInput Size=256x256, #Params=1.7M, #MACs=1.2G, Nano Latency (ms)=22, Mobile Latency (ms)=27, Pi Latency (ms)=1092022.05 | 37.8 | — | — | — | — | — | — | — | — | — | |
| EfficientHRNet-H-4Input Size=384x384, #Params=2.8M, #MACs=2.2G, Nano Latency (ms)=50, Mobile Latency (ms)=78, Pi Latency (ms)=2732022.05 | 35.5 | — | — | — | — | — | — | — | — | — |