Research And Implementation Of Handicap Instruction Recognition For Autonomous Vehicles

This paper describes the research and implementation of unmanned vehicles operating under the command of a vehicle commander.Based on the multi-target detection and key point detection technology to find the face and the hand,this paper completes the face recognition through the comparison of the key points of the face,and compares with the agreed rules according to the position information of multiple parts.Interpret the information expressed by the character.Provide a reference for the next action intent,as well as the 3D modeling representation.The human body key point detection technology combines image processing and analysis,computer vision,artificial intelligence and many other disciplines.It has high use value in military,medical,intelligent monitoring,cultural entertainment and other fields.At present,it can be said that it is widely used.The technology and development prospects are very huge.Based on the in-depth understanding of face recognition algorithm,attitude detection algorithm and gesture recognition algorithm,this paper combines the three algorithms,and then optimizes the algorithm structure to successfully realize the gesture of the authorized person in the case of recognizing the face.Analyze to interpret the meaning of the gesture that the authorizer is to express.The article mainly starts from the traditional rule-based detection algorithm,and the deep learning-based detection algorithm researches the technology of gesture recognition algorithm.Face recognition algorithm.Firstly,the commonly used Haar-like feature extraction method and HOG feature extraction method are analyzed.Secondly,the traditional classifier model is further studied,and the commonly used SVM and BOOST models are mainly studied.Then the face recognition algorithm based on convolutional neural network is studied,and the algorithm is implemented by simulation and applied to the actual hand tracking.Key point detection based on deep learning.Firstly,the basic theory of convolutional neural networks and some classical algorithms in the field are introduced,and some algorithms are analyzed.Then,the analysis focuses on the Open Pose algorithm in the deep learning framework.The Open Pose algorithm has been tested and improved to make it available in our hand tracking.Position and angle based gesture recognition algorithm.The gesture localization algorithm based on face relative position is combined with the prefix tree-based command analysis tree and successfully applied to our gesture recognition algorithm.