Table 4 Compares our method with state-of-the-art methods on the Cambridge datasets.
From: Dynamic gesture recognition based on 2D convolutional neural network and feature fusion
Cambridge | Methods | Top-1 accuracy |
|---|---|---|
Kim et al.10 | Tensor canonical correlation analysis | 82.4% |
Liu et al.32 | Genetic programming | 85.5% |
Lui et al.34 | Tangent bundle | 91.3% |
Wong et al.35 | Probabilistic latent semantic analysis | 91.4% |
Baraldi et al.36 | Dense trajectories + hand segmentation | 94.1% |
Zhao et al.37 | Information theoretic | 96.2% |
Tang et al.33 | Key frames + feature fusion | 98.2% |
Ours | Key frames splicing + feature fusion | 98.6% |
