GESTURE RECOGNITION, GESTURE CONTROL AND NEURAL NETWORK TRAINING METHODS AND APPARATUSES, AND ELECTRONIC DEVICE

摘要

Embodiments of the present application provide gesture recognition, gesture control and neural network training methods and apparatuses, and an electronic device. The gesture recognition method comprises: performing gesture information detection on an image by means of a neural network to obtain a hand possible region in the image, a gesture possible category, and a gesture possible category probability, the gesture possible category comprising no gesture category and at least one gesture category; if the detected gesture possible category of the maximum probability is the no gesture category, not outputting location information of the hand possible region in the image, otherwise outputting the location information of the hand possible region in the image and the detected gesture possible category of the maximum probability. By means of the embodiments of the present application, accurate recognition and tracking of gestures can be achieved.