A New Gesture Recognition Technology Used in Intelligent Inspection Drones

Abstract

The present invention relates to an alarm method for recognizing gestures based on a camera, comprising: setting a gesture data set; establishing a frame by the camera to recognize the trend of human activities; recognizing human gestures by using a convolutional neural network; comparing a captured binary gesture image with a security gesture, and setting a counter parameter with a value of 1; if the gesture image is successfully compared with the security gesture, setting the value of the parameter to plus 1; if comparison between the gesture image and the security gesture fails, setting the value of the parameter to 0; and when the value of the parameter is greater than three, establishing a signal for the camera and the receiving device. The camera of the present invention uses the convolutional neural network to recognize human gesture, processes the human gesture and compares a gesture image with a security gesture. After the conditions are met, the position is sent to a receiving device of the police officer on duty to call the police in time, thereby reducing the frequency of crime. In the whole process, the camera is used for the capturing, and a person injured can effectively prevent more serious persecution without shouting.