List your company

Multi-person 3D Pose Estimation on MuPoTS-3D (test)

89.63DPCKIntegrated Top-Down and Bottom-Up Network
61.769668.994876.2283.4452Jun 15, 2020Aug 8, 2020Oct 1, 2020Nov 24, 2020Jan 17, 2021Mar 12, 2021May 6, 2021

Evaluation results

MethodLinks
Integrated Top-Down and Bottom-Up NetworkGroup=Camera-centric2021.0489.6—————————————————————————————————————————————48———
GnTCNGroup=Camera-centric2020.1287.5—————————————————————————————————————————————45.7———
Cheng et al. [7]Group=Camera-centric2021.0487.5—————————————————————————————————————————————45.7———
TPNTemporal=true, Evaluation protocol=Calculated on detected frames only2020.1086.8100———————————————————————————————————————————225————
TPNTemporal=true, Evaluation protocol=Calculated on all frames2020.1085.3103———————————————————————————————————————————252————
Lin et al. (2020)Group=Camera-centric2020.1283.7—————————————————————————————————————————————35.2———
Lin et al.Group=Camera-centric2021.0483.7—————————————————————————————————————————————35.2———
S3HTBackbone=ResNet-50, Input size=256x2562021.0483.3—————————————————————————————————————————————————
Moon et al. (2019)Group=Camera-centric2020.1282.5—————————————————————————————————————————————31.8———
Moon et al.Group=Camera-centric2021.0482.5—————————————————————————————————————————————31.8———
Li et al. (2020)Group=Camera-centric2020.1282—————————————————————————————————————————————43.8———
Li et al.Group=Camera-centric2021.0482—————————————————————————————————————————————43.8———
Moon et al.Temporal=false, MRPE evaluation=Calculated on detected frames only2020.1081.8————————————————————————————————————————————277————
Moon et al. ICCV'192021.0481.8—————————————————————————————————————————————————
Cheng et al. (2020)Group=Person-centric2020.1280.5—————————————————————————————————————————————————
Zhen et al. (2020)Group=Camera-centric2020.1280.5—————————————————————————————————————————————38.7———
Cheng et al. [8]Group=Person-centric2021.0480.5—————————————————————————————————————————————————
Zhen et al.Group=Camera-centric2021.0480.5—————————————————————————————————————————————38.7———
Veges et al.Temporal=false, MRPE and MPJPE evaluation=Calculated on detected frames only2020.1078.2120———————————————————————————————————————————238————
Cheng et al. (2019)Group=Person-centric2020.1274.6—————————————————————————————————————————————————
Cheng et al. [9]Group=Person-centric2021.0474.6—————————————————————————————————————————————————
BMPTime [s]=0.0562021.0573.83—————————————————————————————————75.34———————————————
Mehta et al. TOG'202021.0472.1—————————————————————————————————————————————————
Benzine et al. CVPR'202021.0472—————————————————————————————————————————————————
Dabral et al. 3DV'192021.0471.3—————————————————————————————————————————————————
LCR-Net++Temporal=false2020.1070.6—————————————————————————————————————————————————
Rogez et al. (2019)Group=Person-centric2020.1270.6—————————————————————————————————————————————————
Rogez et al.Group=Person-centric2021.0470.6—————————————————————————————————————————————————
SMPL R-CNNvariant=full2020.0669.12—————————————————————————————————72.22———————————————
CRMHTime [s]=0.0832021.0569.12—————————————————————————————————72.22———————————————
Rogez et al. TPAMI'192021.0468.9—————————————————————————————————————————————————
BMP-LiteTime [s]=0.0382021.0568.63—————————————————————————————————71.92———————————————
SMPL R-CNNvariant=baseline2020.0666.95—————————————————————————————————68.96———————————————
OpenPose + HMR2020.0666.09—————————————————————————————————70.9———————————————
HMRTime [s]=0.262021.0566.09—————————————————————————————————70.9———————————————
Mask-RCNN + HMR2020.0665.57—————————————————————————————————68.57———————————————
Mehta et al. (2018)Group=Person-centric2020.1265—————————————————————————————————————————————————
Mehta et al.Group=Person-centric2021.0465—————————————————————————————————————————————————
Mehta et al. 3DV'182021.0465—————————————————————————————————————————————————
OpenPose + SMPLify-X2020.0662.84—————————————————————————————————68.04———————————————
SMPLify-XTime [s]=6.42021.0562.84—————————————————————————————————68.04———————————————
3D PoseNet2019.04—292—————————————————————12091—————————————————————————
3DMPPEBounding box and root source=Ground truth2020.10——93.275.680.381.584.675.384.569.390.192818173.473.581.889.688.484.374.570.681.2———————————————————————————
3DMPPEBounding box and root source=Ground truth2020.10——93.98380.381.585.475.384.577.290.192818174.37681.889.688.484.375.576.282.6———————————————————————————
3DMPPE2023.09———————————————————————24——————————————————————31.5—81.843.3
Baseline2019.04—320—————————————————————12291—————————————————————————
BEV2023.09————————————————————————————————————————————————70.2—
DASMethod category=Bottom-up methods, Backbone=2-stage MSPN, Inference device=single V100 GPU, Batch size=12022.03—————————————————————————————————81—————————————75——
DASApproach=Single-stage, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————82.7————————————39.2———
Fabbri et al.Method category=Bottom-up methods2022.03———————————————————————————————————————————————125——
HDNet3D pose estimator=[29]2020.07———————————————————————————————————38.337.836.237.43434.936.429.235.2——————
HMORBackbone=ResNet-50, FPN=true, Flip-test=true, Ensemble=false2020.08—————————————————————————————————82—————————43.843.5—————
JointDepthNetEvaluation Protocol=Comparing every pose, Skeleton Normalization=normalized2020.04——89.575.985.283.98573.483.658.765.190.476.881.96755.980.890.69081.181.168.6———————————78.2————————————————
JointDepthNetEvaluation Protocol=Comparing detected poses only, Skeleton Normalization=normalized2020.04——89.581.685.984.490.573.585.568.965.190.479.182.672.768.1819490.487.490.492.6———————————82.7————————————————
LCRBounding box and root source=Reported2020.10——67.749.853.459.167.522.843.749.931.178.150.25151.649.356.266.565.262.966.159.153.8———————————————————————————
LCRBounding box and root source=Reported2020.10——69.167.354.661.774.525.248.463.36978.153.852.260.560.959.170.5767077.181.462.4———————————————————————————
LCR-netprotocol=Accuracy for person annotations matched to a prediction2017.12—14669.167.354.661.774.525.248.463.36978.153.852.260.560.959.170.5767077.181.462.4———————————————————————————
LCR-netEvaluation Protocol=Overall accuracy, Skeleton=Re-targeted to ground truth bone-lengths2017.12——67.749.853.459.167.522.843.749.931.178.150.25151.649.356.266.565.262.966.159.153.8———————————————————————————
LCR-Net2019.04———————————————————————14686—————————————————————————
LCR++Bounding box and root source=Ground truth2020.10——87.361.967.974.678.848.958.359.778.189.569.273.866.25674.182.178.172.673.16170.6———————————————————————————
LCR++Bounding box and root source=Ground truth2020.10——8873.367.974.681.850.160.660.878.289.570.874.472.864.574.284.985.278.475.874.474———————————————————————————
LCRNet2020.08—————————————————————————————————53.8————————————————
LCRNet++2020.08—————————————————————————————————70.6————————————————
Lin et al.Method category=Top-down methods, reproduced=true2022.03—————————————————————————————————83.7—————————————118——
Lin et al.Approach=Top-down, Evaluation Protocol=matched groundtruths2022.03—————————————————————————————————83.7————————————35.2———
Mehtagroundtruth=all groundtruths used2019.07—————————————————————————62.181.277.957.747.297.366.347.666————————————————
Mehta [20]Evaluation Protocol=Comparing every pose, Skeleton Normalization=normalized2020.04——8160.964.46369.130.36559.664.183.96868.662.359.270.18079.667.366.667.2———————————66————————————————
Mehta [20]Evaluation Protocol=Comparing detected poses only, Skeleton Normalization=normalized2020.04——8165.364.663.97530.365.161.164.183.972.469.97172.971.383.679.673.578.990.9———————————70.8————————————————
Mehta et al.2019.04———————————————————————13293—————————————————————————
Mehta et al. [20]Approach=Bottom-up, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————70.4————————————————
Mehta et al. [21]Approach=Bottom-up, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————65————————————————
Moon [21]Evaluation Protocol=Comparing every pose, Skeleton Normalization=normalized2020.04——94.477.57981.985.372.881.975.790.290.479.279.975.172.781.189.989.681.881.776.2———————————81.8————————————————
Moon [21]Evaluation Protocol=Comparing detected poses only, Skeleton Normalization=normalized2020.04——94.478.67982.186.672.881.975.890.290.479.479.975.3818190.789.683.181.777.3———————————82.5————————————————
Moon et al.2020.08—————————————————————————————————81.8—————————31.539.8—————
Moon et al.Method category=Top-down methods, reproduced=true2022.03—————————————————————————————————81.8—————————————107——
Moon et al.Approach=Top-down, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————81.8————————————31.5———
ORPMprotocol=Accuracy for person annotations matched to a prediction2017.12—132.58164.364.663.773.830.365.160.764.183.971.569.66969.671.182.979.672.276.285.969.8———————————————————————————
ORPMEvaluation Protocol=Overall accuracy2017.12——8159.964.462.86830.36559.264.183.967.268.360.656.569.979.479.666.166.363.565———————————————————————————
Oursgroundtruth=all groundtruths used2019.07—————————————————————————79.192.685.179.46796.685.773.181.8————————————————
PandaNetBounding box and root source=Reported2020.10——————————————————————72———————————————————————————
PI-NetBounding box and root source=Ground truth2020.10——93.577.48282.987.275.98471.590.292.282.582.974.775.783.691.490.68674.971.182.5———————————————————————————
PI-NetBounding box and root source=Ground truth2020.10——93.98581.58388.975.684.77890.492.282.582.67677.683.591.590.585.975.778.583.9———————————————————————————
POTR-3D2023.09———————————————————————10.9——————————————————————50.9—83.716.3
Rogezgroundtruth=all groundtruths used2019.07—————————————————————————49.467.457.151.441.384.656.336.353.8————————————————
Rogez [26]Evaluation Protocol=Comparing every pose, Skeleton Normalization=normalized2020.04——67.749.853.459.167.522.843.749.931.178.150.25151.649.356.266.565.262.966.159.1———————————53.8————————————————
Rogez [26]Evaluation Protocol=Comparing detected poses only, Skeleton Normalization=normalized2020.04——69.167.354.661.774.525.248.463.36978.153.852.260.560.959.170.5767077.181.4———————————62.4————————————————
Rogez [27]Evaluation Protocol=Comparing every pose, Skeleton Normalization=normalized2020.04——87.361.967.974.678.848.958.359.778.189.569.273.866.25674.182.178.172.673.161———————————70.6————————————————
Rogez [27]Evaluation Protocol=Comparing detected poses only, Skeleton Normalization=normalized2020.04——8873.367.974.681.850.160.660.878.289.570.874.472.864.574.284.985.278.475.874.4———————————74————————————————
Rogez et al.Approach=Top-down, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————70.6————————————————
RootNet3D pose estimator=[29]2020.07———————————————————————————————————37.635.63434.130.730.631.325.331.8——————
Single Shot2020.08—————————————————————————————————66————————————————
SingleshotBounding box and root source=Reported2020.10——8160.964.46369.130.36559.664.183.96868.662.359.270.18079.667.366.667.266———————————————————————————
SingleshotBounding box and root source=Reported2020.10——8165.364.663.97530.365.161.164.183.972.469.97172.971.383.679.673.578.990.970.8———————————————————————————
SingleStage2023.09——————————————————————————————————————————————39.3—80.9—
SMAP2023.09——————————————————————————————————————————————35.2—73.5—
TDBU-Netuses GT at testing=true2023.09———————————————————————29.6——————————————————————48—89.6—
VirtualPosereproduced=true2023.09———————————————————————23.1——————————————————————44—72.341.5
Wang et al.Approach=Top-down, Evaluation Protocol=all groundtruths2022.03—————————————————————————————————82————————————43.8———
Xnect2020.08—————————————————————————————————70.4————————————————
XnectBounding box and root source=Reported2020.10——88.465.168.272.576.246.265.864.175.182.474.172.464.458.873.780.484.367.274.367.870.4———————————————————————————