Abstract:
Deep learning algorithms have attracted more attention recently as a result of their unrivalled
ability to automatically learn distinguishing features from enormous amounts of data. Surface
Electromyography(sEMG) signals can be used to extract hidden characteristics, here seven dif ferent gestures are classified by building a deep neural network model. Analysis of the sEMG
can be used to forecast the human’s intended motion. However, there are frequently a lot of
parameters in the models that researchers have lately presented. As a result, the classifica tion accuracy of LSTM model is improved, compared to a simple Convolutional Neural Net work(CNN). The Myo Dataset were used to validate the proposed framework. The gesture
recognition system produced high classification accuracy. The comparison study on numerous
parameters conclude that LSTM model showed the best performance with overall accuracy of
94.78%
