Research On Real-time Target Detection And Tracking System Oriented To Animal Behavior

At present,various mental diseases seriously affect the healthy life of patients.Research behavior can provide great help in the treatment of mental diseases.For the research of various mental diseases,animals are usually used as research models to carry out animal behavior experiments.Animal behavior experiments provide an important basis for the research of mental diseases,and have certain research significance and value.In animal behavior experiments,the collection and recording of experimental data has always been a difficult problem for experimenters.Although manual analysis has been developed into the current image analysis method,this traditional method still has the following shortcomings:(1)Existing Most of the animal behavior analysis software is in the form of offline analysis,which cannot achieve real-time analysis,and the experiment efficiency is low.(2)In traditional animal behavior analysis software,each type of software only supports behavior analysis of a unique scene,and it is not competent in the face of many different animal behavior experimental scenes,and its versatility is low.(3)Most analysis systems use image processing methods,which can only analyze some brief information of animals,such as the position of the mouse's center of gravity,but cannot perform detailed analysis of key points of the animal's body.In order to solve the shortcomings of the traditional animal behavior system,the research work of this paper is to develop a real-time target detection and tracking system for animal behavior.The main research work is as follows:(1)In order to achieve real-time target detection and target tracking,the Retina Face neural network is used to train the model,and NVIDA Tensor Rt is used to accelerate the model and build the engine to realize the forward reasoning of the neural network and improve the reasoning speed of the model.(2)Use Lab VIEW language to write the user operating software of the animal behavior system.In order to solve the problem that the Lab VIEW environment cannot perform deep learning,the C++ program is called in the form of Lab VIEW calling DLL,which realizes the deep learning in the Lab VIEW environment,and the software is easy to operate.And has a high execution efficiency.(3)In order to make the resource utilization of the animal behavior system higher,the parallel processing method is adopted to optimize the program,two frames of data are always cached in the video memory,and the principle of first in,first out is adopted to reduce the time band of data copying between the memory and the video memory.The impact of the future has improved the timeliness of target detection.(4)In order to solve the problem of single application scenario and analysis data of traditional animal behavior system software,a set of animal behavior automatic analysis system that can be applied to a variety of different scenarios is designed.The camera collects images in real time,and the computer analyzes the data in real time.The target detection and tracking of mice,and the analysis of a variety of different behaviors are integrated,and the experimental results are finally saved,reducing the workload of the experimenters and improving the efficiency of the experiment.Set up an animal behavior experiment scene for testing.The system has good detection and tracking effects,and can be applied to a variety of animal behavior experiments,such as open field,water maze and three-box social experiments,and can analyze many different types in real time.The behavioral information of the mouse,such as the trajectory of the mouse,the amount of displacement,the moving speed,the residence time,etc.,meet the needs of animal behavior experiments.