Research Article

Wi-Exercise: An Indoor Human Movement Detection Method Based on Bidirectional LSTM Attention

Table 1

Comparison of recognition results of different methods.

Method namesIdentify actionRecognition rate

KNNSquat-up, sit-up, press-up, lunge, jumping jack, and high knee lift79.87%, 76.57%, 72.96%, 77.97%, 73.78%, and 75.23%
SVMSquat-up, sit-up, press-up, lunge, jumping jack, and high knee lift81.54%, 78.79%, 76.43%, 79.28%, 76.71%, and 77.36%
BPSquat-up, sit-up, press-up, lunge, jumping jack, and high knee lift82.56%, 82.01%, 80.33%, 81.85%, 81.76%, and 81.97%
ABLSTMSquat-up, sit-up, press-up, lunge, jumping jack, and high knee lift96.57%, 95.41%, 92.975%, 94.847%, 92.73%, and 94.015%