List your company
ResearchFormerly WizWand

3D Hand Pose Estimation on ICVL (test)

4.76Mean Error (mm)Virtual View Selection and Fusion Module
4.44726.55868.6710.7814Nov 20, 2017Oct 11, 2018Sep 2, 2019Jul 23, 2020Jun 14, 2021May 5, 2022Mar 27, 2023
Updated 8mo ago

Evaluation results

MethodLinks
Virtual View Selection and Fusion Modulenumber of views=152022.034.76
Virtual View Selection and Fusion Modulenumber of views=92022.034.77
Virtual View Selection and Fusion Modulenumber of views=252022.034.79
Virtual View Selection and Fusion Modulenumber of views=32022.034.86
Virtual View Selection and Fusion Modulenumber of views=12022.035.16
HandFoldingNetPercentage of ground-truth annotations used for training=100%2023.035.95
HandFoldingNetInput=3D, Type=Regression-based (R)2021.085.95
AWR2022.035.98
OursPercentage of ground-truth annotations used for training=100%2023.035.99
OursLabel Usage=100%, Augmented Set=No2023.035.99
JGR-P2OPercentage of ground-truth annotations used for training=100%2023.036.02
JGR-P2OInput=2D, Type=Detection-based (D)2021.086.02
OursLabel Usage=75%, Augmented Set=No2023.036.04
OursLabel Usage=50%, Augmented Set=No2023.036.06
OursPercentage of ground-truth annotations used for training=25%2023.036.11
OursLabel Usage=25%, Augmented Set=No2023.036.11
V2V-PoseNet2017.116.28
V2V2022.036.28
V2V-PoseNetPercentage of ground-truth annotations used for training=100%2023.036.28
V2VInput=3D, Type=Detection-based (D)2021.086.28
V2V*ensemble=10 models, FPS=3.52019.086.286
Point-to-PointPercentage of ground-truth annotations used for training=100%2023.036.3
Point-to-PointInput=3D, Type=Detection-based (D)2021.086.3
P2PFPS=41.82019.086.328
P2P2022.036.33
A2J2022.036.46
A2JPercentage of ground-truth annotations used for training=100%2023.036.46
A2JFPS=105.062019.086.461
NARHTPercentage of ground-truth annotations used for training=100%2023.036.47
CrossInfoNetPercentage of ground-truth annotations used for training=100%2023.036.73
CrossInfoNetInput=2D, Type=Regression-based (R)2021.086.73
Pose-REN2017.116.79
Pose-REN2019.086.79
Pose-RenInput=2D, Type=Regression-based (R)2021.086.79
HandPointNet2022.036.93
HandPointNetFPS=482019.086.935
HandPointNetPercentage of ground-truth annotations used for training=100%2023.036.94
OursLabel Usage=1%, Augmented Set=No2023.036.94
HandPointNetInput=3D, Type=Regression-based (R)2021.086.94
LSPSLabel Usage=100%, Augmented Set=No2023.037
LSPSLabel Usage=75%, Augmented Set=No2023.037.05
LSPSLabel Usage=50%, Augmented Set=No2023.037.1
SHPR-NetPercentage of ground-truth annotations used for training=100%2023.037.22
SHPR-NetInput=3D, Type=Regression-based (R)2021.087.22
DenseReg2022.037.24
DenseRegFPS=27.82019.087.3
DenseRegPercentage of ground-truth annotations used for training=100%2023.037.3
DenseRegInput=2D, Type=Detection-based (D)2021.087.3
REN-9x6x62017.117.31
REN-9x6x62019.087.31
REN-9x6x6Percentage of ground-truth annotations used for training=100%2023.037.31
Ren-9x6x6Input=2D, Type=Regression-based (R)2021.087.31
LSPSLabel Usage=25%, Augmented Set=No2023.037.35
REN-4x6x62017.117.63
REN-4x6x62019.087.63
REN-4x6x6Percentage of ground-truth annotations used for training=100%2023.037.63
Ren-4x6x6Input=2D, Type=Regression-based (R)2021.087.63
SO-HandNetLabel Usage=100%, Augmented Set=No2023.037.7
DeepPrior++2017.118.1
DeepPrior++FPS=302019.088.1
DeepPrior++Percentage of ground-truth annotations used for training=100%2023.038.1
DeepPrior++Input=2D, Type=Regression-based (R)2021.088.1
Baek et al.Label Usage=100%, Augmented Set=Yes, 10 times2023.038.5
Baek et al.(w/o refine)Label Usage=100%, Augmented Set=Yes, 10 times2023.039.1
SO-HandNetLabel Usage=75%, Augmented Set=No2023.039.1
JTSC2017.119.16
JTSC2019.089.16
SO-HandNetLabel Usage=50%, Augmented Set=No2023.039.4
Cascade2017.119.9
Cascade2019.089.9
Crossing NetLabel Usage=50%, Augmented Set=No2023.0310
Crossing NetLabel Usage=75%, Augmented Set=No2023.0310.1
CrossingNets2017.1110.2
CrossingNetsFPS=90.92019.0810.2
Crossing NetLabel Usage=100%, Augmented Set=No2023.0310.2
DeepPrior2017.1110.4
DeepPriorPercentage of ground-truth annotations used for training=100%2023.0310.4
Baek et al.(w/o aug.; refine)Label Usage=100%, Augmented Set=No2023.0310.4
DeepPriorInput=2D, Type=Regression-based (R)2021.0810.4
CDO2017.1110.5
Crossing NetLabel Usage=25%, Augmented Set=No2023.0310.5
Hand3D2017.1110.9
Hand3DFPS=302019.0810.9
SO-HandNetLabel Usage=25%, Augmented Set=No2023.0311.1
DeepModel2017.1111.56
DeepModel2019.0811.56
DeepModelPercentage of ground-truth annotations used for training=100%2023.0311.56
DeepModelInput=2D, Type=Regression-based (R)2021.0811.56
Baek et al.(baseline)Label Usage=100%, Augmented Set=No2023.0312.1
LRF2017.1112.58
LRF2019.0812.58