Design Of Smart Home Control System Based On Human Behavior Recognition Technology

With the development of science and technology,emerging technologies such as artificial intelligence and big data have gradually entered the daily life of people,and smart home products have quickly entered people’s field of vision.In the traditional smart home system,the human behavior recognition in video surveillance mainly relies on labor,which will bring a great workload to the video viewing work.In response to this problem,the paper studies human behavior recognition technology and applies human behavior technology to the smart home system to realize the function of automatically classifying video behavior.Thus,a smarter and more convenient smart home system has been established.The main work done in this paper includes the following contents:Firstly,it analyzes the research background and significance of smart home and human behavior recognition,and analyzes the development and research status of smart home and human behavior recognition at home and abroad.According to the functional requirements of the system,the overall analysis and design of the system is carried out,and the overall framework of the system consisting of PC,main control center and control node is determined.The PC end is mainly responsible for the processing of video data,and the main control center is responsible for controlling the transmission of each control node and video data,and the control node is responsible for controlling each household electrical appliance.On this basis,the hardware and software design of the system was carried out.Secondly,the human behavior recognition algorithm is studied in detail.Several common human behavior recognition algorithms based on single-stream convolutional networks,3D CNN-based networks and two stream convolutional networks are introduced,and their shortcomings are analyzed.Then,Then improved the two stream convolutional neural network.Based on the two stream convolutional network,LSTM is combined to avoid the lack of long-term information in the feature information of the two stream convolutional network.Therefore,the problem that the two stream convolution network uses the sample frame as the input data causes the tag information to be incomplete or even missing is solved.Experimental analysis of the improved algorithm shows that the improved algorithm has a higher recognition rate.Then the smart home system based on human behavior recognition is implemented again,including the realization of video data transmission in the main control center,the software implementation of PC-side behavior recognition and the realization of home appliance control.Finally,the paper tests the system.The test results show that the smart home system designed in this paper can control the home appliances in each room,and at the same time can effectively identify human behavior in video images.