We are happy to include your results, please send your predictions on the test set to [eldar at mpi-inf.mpg.de]

Leeds Sport Pose (LSP) Dataset

Evaluation toolkit

 

PCP evaluation measure

Observer-Centric (OC) annotations

MethodTorsoUpper legLower legUpper armForearmHeadPCP
Kiefel&Gehler, ECCV'14, (predictions)84.374.567.654.128.378.361.2
Pishchulin et al., CVPR'13, (predictions)87.475.768.054.433.777.462.8
Fu et al., ICCV'15, (predictions)85.475.072.062.048.077.767.7
Ramakrishna et al., ECCV'14, (predictions)88.179.073.662.839.580.467.8
Ouyang et al., CVPR'14, (predictions)88.677.871.961.945.484.368.7
Pishchulin et al., ICCV'13, (predictions)88.778.973.261.845.085.169.2
Chen&Yuille, NIPS'14, (predictions)92.782.977.069.255.487.875.0
Yang et al., CVPR'16, (predictions)96.588.781.778.866.783.181.1
Chu et al., CVPR'16, (predictions)95.487.683.276.965.289.681.1
Pishchulin et al., CVPR'16, (predictions)*96.091.083.582.871.896.285.0

* methods trained when adding MPII training set to the LSP training and LSP extended training set

Table as TEX

 

Person-Centric (PC) annotations

MethodTorsoUpper legLower legUpper armForearmHeadPCP
Wang&Li, CVPR'13, (predictions)87.556.055.843.132.179.154.1
Pishchulin et al., ICCV'13, (predictions)88.763.658.446.035.285.158.0
Tompson et al., NIPS'14, (predictions)90.370.461.163.051.283.766.6
Fan et al., CVPR'15, (predictions)95.477.769.862.849.186.670.1
Carreira et al., CVPR'16, (predictions)95.381.873.366.751.084.472.5
Chen&Yuille, NIPS'14, (predictions)96.077.272.269.758.185.673.6
Yang et al., CVPR'16, (predictions)95.678.571.872.261.883.974.8
Rafi et al., BMVC'16, (predictions)97.687.380.276.866.293.381.2
Belagiannis&Zisserman, arXiv'16, (predictions)*96.086.782.279.469.489.482.1
Lifshitz et al., ECCV'16, (predictions)**97.388.884.480.671.494.884.3
Pishchulin et al., CVPR'16, (predictions)*97.088.882.082.471.895.884.3
Yu et al., ECCV'16, (predictions)98.093.188.182.972.683.085.4
Insafutdinov et al., ECCV'16, (predictions)*97.090.686.986.179.595.487.8
Wei et al., CVPR'16, (predictions)*98.082.289.185.877.995.088.3
Bulat&Tzimiropoulos, ECCV'16, (predictions)**97.792.489.386.779.795.288.9
Chu et al., CVPR'17, (predictions)*98.495.092.888.581.295.790.9
Ning et al., TMM'17, (predictions)*98.695.893.690.784.296.492.3
Yang et al., ICCV'17, (predictions)*99.195.793.991.184.396.792.6
Chou et al., arXiv'17, (predictions)*99.196.094.290.485.296.892.8

* methods trained when adding MPII training set to the LSP training and LSP extended training set

** methods trained when adding MPII training set to the LSP training set

Table as TEX

 

PCK evaluation measure

Note: no ground truth information was used to generate predictions by each method.

 

Observer-Centric (OC) annotations

PCK @ 0.2
MethodHeadShoulderElbowWristHipKneeAnklePCK
Kiefel&Gehler, ECCV'14, (predictions)83.573.755.936.273.770.566.965.8
Ramakrishna et al., ECCV'14, (predictions)84.977.861.447.273.669.168.869.0
Fu et al., ICCV'15, (predictions)81.876.663.052.873.670.867.169.4
Ouyang et al., CVPR'14, (predictions)86.578.261.749.376.970.067.670.0
Pishchulin et al., ICCV'13, (predictions)87.577.661.447.679.075.268.471.0
Chen&Yuille, NIPS'14, (predictions)91.584.770.363.282.778.172.077.5
Yang et al., CVPR'16, (predictions)90.689.180.373.585.582.868.881.5
Chu et al., CVPR'16, (predictions)93.787.278.273.888.283.080.983.6
Pishchulin et al., CVPR'16, (predictions)*97.492.083.879.093.188.383.788.2

* methods trained when adding MPII training set to the LSP training and LSP extended training set

Table as TEX

 

Click on plots for high-resolution versions

 

   
   
   

 

Person-Centric (PC) annotations

PCK @ 0.2
MethodHeadShoulderElbowWristHipKneeAnklePCK
Wang&Li, CVPR'13, (predictions)84.757.143.736.756.752.450.854.6
Pishchulin et al., ICCV'13, (predictions)87.256.746.738.061.057.552.757.1
Tompson et al., NIPS'14, (predictions)90.679.267.963.469.571.064.272.3
Fan et al., CVPR'15, (predictions)92.475.265.364.075.768.370.473.0
Carreira et al., CVPR'16, (predictions)90.581.865.859.881.670.662.073.1
Chen&Yuille, NIPS'14, (predictions)91.878.271.865.573.370.263.473.4
Yang et al., CVPR'16, (predictions)90.678.173.868.874.869.958.973.6
Rafi et al., BMVC'16, (predictions)95.886.279.375.086.683.879.883.8
Yu et al., ECCV'16, (predictions)87.288.282.476.391.485.878.784.3
Belagiannis&Zisserman, arXiv'16, (predictions)*95.289.081.577.083.787.082.885.2
Lifshitz et al., ECCV'16, (predictions)**96.889.082.779.190.986.082.586.7
Pishchulin et al., CVPR'16, (predictions)*97.091.083.878.191.086.782.087.1
Insafutdinov et al., ECCV'16, (predictions)*97.492.787.584.491.589.987.290.1
Wei et al., CVPR'16, (predictions)*97.892.587.083.991.590.889.990.5
Bulat&Tzimiropoulos, ECCV'16, (predictions)**97.292.188.185.292.291.488.790.7
Chu et al., CVPR'17, (predictions)*98.193.789.386.993.494.092.592.6
Yang et al., ICCV'17, (predictions)*98.394.592.288.994.795.093.793.9
Ning et al., TMM'17, (predictions)*98.294.491.889.394.795.093.593.9
Chou et al., arXiv'17, (predictions)*98.294.992.289.594.295.094.194.0

* methods trained when adding MPII training set to the LSP training and LSP extended training set

** methods trained when adding MPII training set to the LSP training set

Table as TEX

 

Click on plots for high-resolution versions