A Dense-Sparse Complementary Network for Human Action Recognition based on RGB and Skeleton Modalities

Open paper
Benchmark Model Rank Results
action-recognition-in-videos-on-ntu-rgbd-120DSCNet (RGB + Pose)#1Accuracy (Cross-Setup): 96.7Accuracy (Cross-Subject): 95.6
action-recognition-in-videos-on-ntu-rgbd-60DSCNet (RGB + Pose)#1Accuracy (CS): 97.4Accuracy (CV): 99.4
action-recognition-in-videos-on-pku-mmdDSCNet (RGB + Pose)#2X-Sub: 97.4X-View: 98.8
skeleton-based-action-recognition-on-n-uclaDSCNet (RGB + Pose)#1Accuracy: 99.1
skeleton-based-action-recognition-on-uavDSCNet (RGB + Pose)#2CSv1(%): 47.3CSv2(%): 71.1

Rank counts only results with a code link.