List your company

3D Human Pose Estimation on Human3.6M v1 (test)

71.9Avg PerformancePavlakos et al.
18.75632.55346.3560.147Nov 28, 2018Jul 10, 2019Feb 20, 2020Oct 2, 2020May 14, 2021Dec 25, 2021Aug 7, 2022

Evaluation results

MethodLinks
Pavlakos et al.Ground-truth Bounding Boxes=true2018.1171.967.471.966.769.172776568.383.796.571.765.874.959.163.2
Tekin et al.2018.1169.754.261.460.261.279.478.363.181.670.1107.369.370.374.351.863.2
Martinez et al.Ground-truth Bounding Boxes=true2018.1162.951.856.258.15969.578.455.258.17494.662.359.165.149.552.4
Martinez et al.Input type=Monocular2019.0562.951.856.258.15969.578.455.258.17494.662.359.165.149.552.4
Fang et al.2018.1160.450.154.35757.166.673.353.455.772.888.660.357.762.747.550.6
Sun et al.Extra Data=true2018.1159.152.854.854.254.361.867.253.153.671.786.761.553.461.647.153.4
Yang et al.Extra Data=true2018.1158.651.558.950.45762.165.449.852.769.285.257.458.443.660.147.7
Hossain & LittleTemporal Information=true, Ground-truth Bounding Boxes=true2018.1158.348.450.757.255.263.172.65351.766.180.95957.362.446.649.6
Hossain & LittleInput type=Monocular, Temporal information usage=true2019.0558.348.450.757.255.263.172.65351.766.180.95957.362.446.649.6
Multi-View MartinezInput type=Multi-view2019.055746.548.65451.567.570.748.549.169.879.457.853.156.742.245.4
Pavlakos et al.Input type=Multi-view2019.0556.941.249.242.843.455.646.940.363.797.611952.142.751.941.839.4
Pavlakos et al.Extra Data=true2018.1156.248.554.454.45259.465.349.952.965.871.156.652.960.944.747.8
Sharma et al.Number of pose predictions=5, Oracle protocol=true, 2D input detector=Stacked Hourglass, Note=estimated from figure2020.1053.4———————————————
Luvizon et al.Ground-truth Bounding Boxes=true, Extra Data=true2018.1153.249.251.647.650.551.860.348.551.761.570.953.748.957.944.448.9
PoseAugVenue=CVPR'21, Temporal Frames (f)=1, 2D Detector Type=CPN2022.0352.9———————————————
Lee et al.Temporal Information=true, Ground-truth Bounding Boxes=true2018.1152.840.249.247.852.650.17550.24355.873.954.155.658.243.343.3
Tome et al.Input type=Multi-view2019.0552.843.349.64248.851.164.340.343.36695.250.252.251.143.945.3
Li et al.Number of pose predictions=5, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1052.743.848.649.149.857.661.545.948.36273.454.850.65643.445.5
Li et al.Number of pose predictions=8, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1052.6———————————————
Dabral et al.Venue=ECCV'18, Temporal Frames (f)=12022.0352.144.850.444.74952.961.443.545.563.187.351.748.552.237.641.9
PoseFormerVenue=ICCV'21, Temporal Frames (f)=9, 2D Detector Type=CPN2022.035247.951.549.350.853.758.649.546.66270.352.649.353.840.543
GraphMDNNumber of pose predictions=5, Oracle protocol=true, 2D input detector=Stacked Hourglass, Model variant=Wide2020.1051.944.248.747.249.455.16244.846.959.776.553.248.953.644.544.4
VideoPose3DTemporal Information=false, Number of Frames=12018.1151.847.150.64951.853.661.449.447.459.367.452.449.555.339.542.7
GraphMDNNumber of pose predictions=8, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1050.943.447.446.348.155.259.343.945.858.675.252.548.253.242.243.6
VideoposeVenue=CVPR'19, Temporal Frames (f)=9, 2D Detector Type=CPN2022.0350.646.448.945.749.852.561.547.746.859.968.150.747.552.738.442.1
Volumetric single viewInput type=Monocular, Approximate pelvis estimation=true2019.0549.941.949.246.947.650.757.941.250.957.374.948.644.341.352.842.7
Ray3DTemporal Frames (f)=9, 2D Detector Type=CPN2022.0349.744.748.748.748.45159.946.846.958.761.750.246.451.538.641.8
Sun et al.Input type=Monocular2019.0549.6———————————————
Kadkhodamohammadi & PadoyInput type=Multi-view2019.0549.139.446.94142.753.654.841.45059.978.849.846.251.140.541
VideoPose3DTemporal Information=true, Number of Frames=243, Convolution Type=causal2018.114945.948.544.347.851.957.846.245.659.968.550.646.45134.535.4
Cai et al.Venue=ICCV'19, Temporal Frames (f)=72022.0348.844.647.445.648.850.85947.243.957.961.949.746.651.337.139.4
RIEVenue=ACMMM'21, Temporal Frames (f)=9, 2D Detector Type=CPN2022.0348.644.847.946.147.450.457.645.744.65764.249.545.750.936.639.8
VideoPose3DTemporal Information=true, Number of Frames=243, Convolution Type=full, Ground-truth Bounding Boxes=true2018.1147.145.147.4424649.156.744.544.457.266.147.544.849.232.634
VideoPose3DTemporal Information=true, Number of Frames=243, Convolution Type=full2018.1146.845.246.743.345.648.155.144.644.357.365.847.1444932.833.9
Pavllo et al.Input type=Monocular, Temporal information usage=true2019.0546.845.246.743.345.648.155.144.644.357.365.847.1444932.833.9
Sharma et al.Number of pose predictions=200, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1046.837.843.24344.351.157.139.74356.36448.145.450.437.939.9
Martinez et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.0846.845.246.743.345.648.155.144.644.357.365.847.1444932.833.9
Yeh et al.Venue=NIPS'19, Temporal Frames (f)=12022.0346.744.846.143.346.44955.244.64458.362.747.143.948.632.733.3
GraphMDNNumber of pose predictions=200, Oracle protocol=true, 2D input detector=Stacked Hourglass, Model variant=Wide2020.1046.24043.24143.45053.640.141.452.667.348.144.24939.540.2
UGCNVenue=ECCV'20, Temporal Frames (f)=962022.0345.641.343.94442.24857.142.243.257.361.34743.54732.631.8
Zhao et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.0843.837.849.437.640.945.141.440.148.350.142.253.544.340.547.339
Li et al.Number of pose predictions=5, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1042.635.539.841.342.34648.936.937.35160.644.940.244.133.136.9
GraphMDNNumber of pose predictions=5, Oracle protocol=true, 2D input detector=Stacked Hourglass, Model variant=Wide2020.1040.533.838.738.339.443.647.834.235.248.558.642.737.841.433.534.4
GraphMDNNumber of pose predictions=8, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1040.133.838.337.838.843.846.333.934.647.958.642.437.641.13234.2
Zhou et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=true2022.0839.934.442.436.642.138.239.834.740.245.660.83942.64229.831.7
Liu et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.0837.836.840.33336.337.54539.734.940.347.737.438.538.629.632
GraphMDNNumber of pose predictions=200, Oracle protocol=true, 2D input detector=Stacked Hourglass, Model variant=Wide2020.1036.330.834.733.634.239.642.23131.942.953.538.134.13829.631.1
Ci et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=true2022.0836.336.338.829.737.834.642.539.832.536.239.534.438.438.231.334.2
Xu et al.Input=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.0835.835.838.13135.335.843.237.331.738.445.535.436.736.827.930.7
Refinement TransformerInput=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.0834.232.237.429.733.433.438.138.132.137.140.133.934.135.228.628.8
JointformerInput=GT keypoints, Protocol=Protocol #1, Extra Data=false2022.08343136.630.233.433.53937.131.337.140.133.833.53528.729.1
Sharma et al.Number of pose predictions=200, Oracle protocol=true, 2D input detector=Stacked Hourglass2020.1032.727.627.534.932.333.342.728.72836.142.73630.737.624.327.1
RANSAC (baseline implementation)Input type=Multi-view2019.0527.424.126.12424.6272523.326.831.449.527.825.42427.424.1
AlgebraicInput type=Multi-view, Confidence usage=without confidence2019.0526.922.925.323.72329.225.12126.234.141.929.223.322.326.623.3
AlgebraicInput type=Multi-view2019.0522.620.422.620.519.722.120.619.52325.8332321.620.723.721.3
VolumetricInput type=Multi-view, Aggregation strategy=sum aggregation2019.0521.319.320.520.119.320.619.81922.923.529.82221.419.822.120.3
VolumetricInput type=Multi-view, Aggregation strategy=softmax aggregation2019.0520.818.82019.318.720.219.318.722.323.329.121.220.319.321.619.8
VolumetricInput type=Multi-view, Aggregation strategy=conf aggregation2019.0520.819.92018.918.520.519.418.422.122.528.721.220.819.722.120.2