EURASIP Journal on Image and Video Processing

Table 3 Results on DHG-14/28 Dataset

From: Spatial temporal graph convolutional networks for skeleton-based dynamic hand gesture recognition

Method	14 gestures	28 gestures
SoCJ+HoHD+HoWR [16]	83.1	80.0
De Smedt et al. [6]	82.5	68.1
CNN + LSTM [9]	85.6	81.1
Chen et al. [11]	84.6	80.3
DPTC [18]	85.8	80.2
STA-Res-TCN [12]	89.2	85.0
HG-GCN	89.2	85.3

The second column describes the accuracy rates of 14 gestures setting. The last column describes the accuracy rates of 28 gestures setting

Back to article page