TRAINING METHOD FOR GESTURE RECOGNITION MODEL, GESTURE RECOGNITION METHOD, AND APPARATUS

摘要

A training method for a gesture recognition model, a gesture recognition method, and an apparatus. The training method for a gesture recognition model comprises: obtaining a training set, the training set comprising a plurality of gesture sample images, a plurality of gesture types being included among the plurality of gesture sample images (11); obtaining a hybrid model, the hybrid model comprising a plurality of gesture recognition models (12); obtaining, for a plurality of the gesture sample images, prediction scores with respect to each gesture type from each of the hand gesture recognition models (13); determining, according to the prediction scores, a target weight combination for the hybrid model from a plurality of weight combinations to be trained, wherein each weight combination among the plurality of weight combinations to be trained comprises a weight corresponding to each gesture recognition model (14). The present method increases recognition accuracy of a gesture recognition model.