PROCEDURES AND PREPARATION FOR TRAINING A MODEL FOR THE RECOGNITION OF HAND AND HAND CLOSURE POINTS AND PREPARATION FOR THE RECOGNITION OF HAND CLOSURE POINTS

摘要

A method for training a model for recognizing key points of a hand, comprising: converting a virtual image of a sample into a simulated image by means of a Cycle-GAN model, wherein the virtual image of the sample is an image generated by means of 3D modeling and comprises coordinates of key points corresponding to key points of a hand, and the simulated image simulates an image acquired in a real scene; extracting an image of the hand from the simulated image; and according to the image of the hand in the simulated image and the coordinates of the key points, training a model for recognizing key points of a hand, wherein the model for recognizing key points of a hand is used to output, according to an input real image, coordinates of key points of a hand in the real image.