Table 2.
YEAR | SUMMARY | DATASET | |
---|---|---|---|
Yang et al. [98] | 2012 | Use of Depth Motion Maps (DMM), combining them with HOG descriptors. | MSRAction3D [107] |
Oreifej et al. [99] | 2013 | Use of histogram of oriented 4D surface normals (HON4D) descriptor. | MSRAction3D [107] MSRGesture3D [108] 3D Action Pairs [99] |
Liu et al. [102] | 2018 | Use of a two-layer BoVW model, using motion-based and shape-based STIPs to distinguish the action. | MSRAction3D [107] UTKinect-Action [109] MSRGesture3D [108] MSRDailyActivity3D [100] |
Satyamurthi et al. [104] | 2018 | Use of multi-directional projected depth motion maps (MPDMM). | MSRAction3D [107] MSRGesture3D [108] |