ResNet-Fused External Attention Network with Fractional Star Piranha Forager Optimization for Forest Fire Detection in Internet of Things Environments

A forest fire is an uncontrolled fire that occurs in natural areas and causes extensive damage to human lives. Therefore, the ResNet-fused External Attention Network with Fractional Star Piranha Forager Optimization (RfEANetFStPFO) is devised for forest fire detection. Initially, the IoT simulation is performed in a cloud-based environment and routing is done by Star Piranha Forager Optimization (StPFO). StPFO combines Piranha Foraging Optimization Algorithm (PFOA) and Star Fish Optimization Algorithm (SFOA). Besides, the forest fire images are collected from the relevant dataset. These collected images are filtered, and several features are extracted for further processing. Subsequently, the forest fire is detected by ResNet-fused External Attention Network (ResfEANet) and its parameters are tuned by the proposed Fractional Star Piranha Forager Optimization (FStPFO). FStPFO integrates the benefits of fractional calculus and StPFO. The RfEANetFStPFO attained an accuracy of 96. 490%, True Positive Rate (TPR) of 97. 691%, and False Positive Rate (FPR) of 2. 869% for 90% of training samples. The routing algorithm named as StPFO obtained an end-to-end delay of 0. 256Formula: see texts (seconds), network life of 0. 756, throughput of 97. 249 Mbps (Megabits per second), and energy consumption of 3. 735 Joules (J) for 1000 rounds with Forest Fire dataset from Mendeley.