Fire Detection Using a Dynamically Developed Neural Network

Early warning systems are critical in providing emergency response in the event of unexpected hazards. Cheap cameras and improvements in memory and computing power have enabled the design of fire detectors using video surveillance systems. This is critical in scenarios where traditional smoke detectors cannot be installed. In such scenarios, it has been observed that the smoke is visible well before flames can be sighted. This paper proposes a method to detect fire flame and/or smoke in real-time by processing the video data generated by ordinary camera monitoring a scene. The objective of this work is recognizing and modeling fire shape evolution in stochastic visual phenomenon. It focuses on detection of fire in image sequences by applying a hybrid algorithm that depends on optimizing the structure of a feed forward neural network. Fire detection experiments using various algorithms were carried. Results show that the proposed algorithm is very successful in detecting fire and/or smoke.