Gesture recognition method and system based on multi-channel information fusion

The invention relates to the field of gesture intelligent recognition, and particularly discloses a gesture recognition method and system based on multi-channel information fusion, and the method comprises the steps: extracting local implicit high-dimensional features of a plurality of surface electromyogram signals of the surface of an arm of a subject through employing a convolutional neural network model of a double attention mechanism; the channel attention and the space attention respectively pay attention to the feature information of the feature content and the feature position in the image, so that mutual complementation is realized to a certain extent, and the feature extraction effect of the network is improved. In the feature fusion process, information compensation is carried out through displacement of probability distribution, information balance between feature vectors can be achieved through compensation information degradation, and then the classification accuracy is improved. In this way, th