The use of ‘Internet of Things’ in homes has grown rapidly over the past few years. However, these systems have always been operated via a remote control, voice control or through user applications. In this paper, we are proposing a system design that provides a way for identifying a wide range of human emotions by employing a hybrid deep learning CNN algorithm to index the emotional state and send the same over a ZigBee wireless module to an Arduino. In order to reflect the user's sentiment, the music and lighting are altered by the Arduino. The system is also equipped with an LCD device that displays the current emotion and room temperature as measured by the temperature sensor.

Keywords : Home Automation, Neural Networks, CNN-LSTM, Facial expression recognition, IoT, Arduino, Zigbee