Published Paper Details:

In recent years, fire accidents mainly in buildings are considerably increasing around the world. The reasons being due to overheating of pots and pan, improper way of keeping portable heaters, smoking in bedrooms, electrical equipment, candles near to explosive or flammable materials, faulty wiring, congested construction of the building and even due to increasing the number of population as well as improper light fittings. Fire outbreak is the third biggest risk to smart cities in India. According to the Indian Risk Survey (IRS) in 2018 and 2019, 7.24% of accidents occur due to fire outbreaks [1]. So the demand for an automation system to detect fire and take appropriate actions such as giving alerts as well as taking immediate action to reduce the intensity of fire has become obvious. In this zest, mainly two approaches are experimented and explored here. Primarily fire detection using machine learning approaches is achieved and also rule-based approach is employed with other relevant parameters. Different machine learning approaches like Deep Learning Neural Network (DNN), AlexNet, VGG-16, LeNet-5, and ResNet-50 experimented for classification of images and detection of fire based on image dataset. Also for rule-based system input from the context-aware sensor system was taken and adjudged the intensity of fire according to different rules to initiate appropriate actions.