Design and Evaluation of an Electronic Eye for Fire Detection in Human Space Capsule

Abstract

Fire is often considered as a good friend, as it helps human in several ways, but the same is also considered as a dangerous foe, once it gets out of control. Fire can burn almost anything in no time and also depletes oxygen from the surrounding atmosphere, thus leading to casualties due to lack of oxygen, smoke and suffocation. Hence there is an urge to design early fire detection systems that can sense the initiation of fire and in turn activate the extinguishers to extinguish the fire, reducing the loss of property and life. In this paper an attempt is made to design and evaluate Electronic eye (E-Eye) for early fire detection in a human space capsule. Two variants of E-eye are proposed and their performances are compared. An attempt is also made to employ the concepts of machine learning using Generative Adversarial Network (GAN) models for fire detection. Maximum recognition accuracy of 100% with a prediction time of around 80ms was achieved using the proposed model. The proposed work is an outcome of a funded project from Indian Space Research Organization. The proposed prototypes can be easily employed in other places too such as houses, offices, storeroom, garage, etc.