Enhanced Fire Detection for Industrial and Environmental Safety Using IoT-Integrated Vision Sensors and Residual Clutch Attention Network and Depthwise Distinguishable Convolutional Neural Networks

Fire detection is an important factor in improved safety in industrial and environmental conditions. The conventional flame detectors, largely relying on heat and smoke detectors, have drawbacks of time lag, low sensitivity, and low coverage when compared with our requirement in large industrial scenes and vast natural terrains vulnerable to fire occurrence. These systems are highly unsuitable for dynamic and high-risk situations as they require a higher accuracy and a faster reaction time for the alarms to be efficient as is now the case due to the aforementioned reasons. In an attempt to remedy these limitations, this research introduces a fire detection system using IoT and vision sensors that combines a RCAN with DDCNN. This architecture is tuned using the Crayfish Optimization Algorithm (COA), which seeks to improve model parameter to improve response time and accuracy. By incorporating feedbacks from sensors placed at its different parts and reading from the environment, the RCAN-DDCNN model can quickly identify the fire patterns and respond by activating the sprinkler systems as well as setting off alarms. The proposed system presents a superior option for fire detecting because it is scalable and adaptive and it provides high degree of precision, making it an optimal development in fire detection technology for industry and environment. The introduced approach attains higher accuracy as 99%.