Dynamic Sign Language Recognition Based on CBAM with Autoencoder Time Series Neural Network

Table 2

A comparison of the final results of the proposed method with other methods on SLR dataset.

Network	Top accuracy (%)
C3D (RGB)	70.53
CBAM-C3D (RGB)	71.77
CBAM-C3D (RGB + optical)	72.86
3D-ResNet (RGB)	79.52
ResNet-LSTM (RGB-base on autoencoder)	82.34
CBAM-3D-ResNet (RGB + optical)	83.96
I3D (RGB) [32]	84.56
CBAM-I3D (RGB) [32]	86.00
I3D (RGB + optical) [32]	88.18
CBAM-ResNet-Bi-LSTM (RGB-base on autoencoder)	89.90
CBAM-I3D (RGB + optical) [32]	90.76