Gesture identification method and system of robot, and robot

Abstract

The invention discloses a gesture identification method and system of a robot, and the robot. The method comprises steps of pre-acquiring pictures which contain different gestures and pictures which do not contain gestures so as to obtain a sample picture set; according to the sample picture set, making a detection sample set and a filtering sample set; according to the detection sample set, carrying out training so as to obtain an adaboost cascading gesture detector; according to filtering sample set, carrying out training so as to obtain a gesture identification convolution neural network; and identifying the acquired gesture pictures through the adaboost cascading gesture detector so as to obtain gesture identification results, and filtering the gesture identification results through the gesture identification convolution neural network so as to obtain a correct gesture identification result. According to the invention, by filtering the gesture identification results through the gesture identification convolution neural network, a gesture can be precisely identified under the complex background.

Description

technical field [0001] The invention relates to the fields of artificial intelligence and picture processing, in particular to a robot gesture recognition method, system and robot. Background technique [0002] With the development of science and technology, intelligent robots are used more and more in daily life and industrial production. In the process of realizing robot intelligence, gesture recognition is an important way of human-computer interaction, and its research and development affect the naturalness and flexibility of human-computer interaction. [0003] At present, there are many service robots that can recognize user instructions according to user gestures and help people complete many tasks. The process of conventional image processing technology and machine learning method to realize gesture recognition usually includes steps such as gesture segmentation, gesture analysis and gesture recognition. This method is usually suitable for recognition in a single b...

AI Technical Summary

Problems solved by technology

This method is usually suitable for recognition in a single background. However, in real applications, gestures are usually in complex environments, such as complex backgrounds, too bright or too dark lights, and different distances between gestures and acquisition devices; in complex environments In this situation, the machine learning method is prone to misjudgment. At this time, manual screening is required, which cannot meet the purpose of intelligent detection.

Images