Research On Intelligent Emotion Lamp Based On Face Expression Recognition

Lamps are indispensable lighting tools in daily’s life and work.With the continuous improvement of living standards,people have put forward higher and higher requirements for lamps.Consequently,intelligent lamps come into being.Although people can find many intelligent lamps on the market with functions such as mobile APP remote control,custom light color,multiple scene modes and voice control,it is still difficult to find an intelligent lamp that can interact with people by automatically recognizing facial expressions.As a result,this thesis designs and develops an intelligent emotion lamp system based on facial expression recognition.In order to realize the facial expression recognition function,the intelligent emotion lamp system first detects the face area in the picture through the Haar cascade classifier,and then obtains the expression recognition result of the face area through the facial expression recognition algorithm,and finally lights up different colors of lights and plays different types of music according to the recognition result to help people ease their emotions and release stress.In order to improve the recognition accuracy and robustness of facial expression recognition algorithms,this thesis combines deep learning and integrated learning for facial expression recognition.The facial expression recognition algorithm combines three convolutional neural network models with different structures into an integrated network model through integrated learning,and assigns weights to the three sub-network models through the weighted voting method,so that the sub-network model with high recognition accuracy on the validation set has a large weight and with low recognition accuracy on the validation set has a small weight.The system consists of an intelligent emotion lamp,a PC,the Alibaba Cloud IoT(Internet of Things)and a mobile phone.With the shape of a cylinder-like body composed of a sound hole,a transparent white shell and a base,the intelligent emotion lamp,the most important part of the system,consists of a camera,a raspberry Pi development board,a color light module and an audio module.The main function of the PC is to train the facial expression recognition model offline,and send the facial expression recognition results to the intelligent emotion lamp online.Alibaba Cloud IoT can enable intelligent emotion lamp connect to the Internet and deliver data to the intelligent emotion lamp according to the rules engine.After the bluetooth of the mobile phone is successfully connected with the intelligent emotion lamp,the songs of the mobile music APP can be played in the intelligent emotion lamp.The WeChat applet ofthe mobile phone can remotely control the color,the brightness and the switch of the intelligent emotion lamp through the Alibaba Cloud IoT.To verify the performance of the proposed system,several typical experiments are conducted.The test results show that the intelligent emotion lamp can detect the human face,in most cases,the PC can recognize the emotion of the facial image sent by the intelligent emotion lamp,then the intelligent emotion lamp quickly lights the color and sounds the background music after receiving the facial expression recognition result sent by the PC;the intelligent emotion lamp can play the songs in the mobile music APP;the WeChat applet can remotely select the light color,adjust the brightness of the lamp,and control the switch.