Many subsystems, including those for lighting, security, weather management, entertainment etc., are managed and coordinated by a central control system in smart homes. By gathering information from multiple sensors, the proposed method improves the automation system remotely using a WiFi-enabled environment. Contours are used to categorize objects and people in the surroundings. Performance is measured by analyzing the IDI datasets (IoT Device identification) and controlled. The main objective of this system is to behave system intelligently. This system can achieve low energy consumption and excellent computational performance, when the devices are increased. The challenges faced in the existing system : high energy consumption, large installation footprint, and subpar effectiveness. It provides the accuracy of 75% to 94% based on the parameters :Pattern, light, Temp, Humidity, Motion etc., using machine learning techniques. Our proposed system, training a model to recognise the power consumption of the devices, categorize and acquire data from the environment using SGDClassifier algorithm. The major objective of this system is to classify the devices and control them based on the high and low power consumption. Performance analysis of algorithms and epochs are discussed and experimental results show the system accuracy from 95.89%- 96.75% based on Human Recognition Percentage(HRP), Watt, Epoch and learning rate parameter and achieves less consumption time with low computational energy.

IoT, IDI dataset, SGDClassifier, Accuracy.