Abstract: A coincide of technologies in Artificial Intelligence and inescapable computing as well as the development of powerful sensors and actuators has gotten interest in the development of smart mediums to transpire and uphold important functions in Daily Living Activities (ADLs). This system proposes an intelligent system for home automation using Internet of things. First, we propose a model using fine tune neural network algorithms. This fine tune neural network has to do with transfer learning. By transfer learning, we mean transferring a knowledge of an existing model into the new model for faster training and for better training performance. This transfer learning was implemented in python by importing mobileNetv2 from keras.applications using tensorflow framework. Secondly, we proposed an emotion expression dataset from which our proposed model will learn to recognize the emotion of a person in the home by means of facial recognition. The model will be able to tell how the person feels, if the person is happy, sad, angry and so on. From the person’s emotional expression, the model will be able to automate any home appliances according to how the person feels. The emotional dataset used in our work is the FER 2013 dataset, which was downloaded from kaggle.com. After successfully training, we had an accuracy of about 88% on all the classes, and had an accuracy of 97% on two of the image classes. Our trained model was saved and exported to web using python flask, were we carried out our testing on a live we camera video.

Keywords: Fine tune Neural Network, Home Automation, Internet of Things, Home Appliances

PDF | DOI: 10.17148/IJARCCE.2021.10204

Open chat
Chat with IJARCCE