Evolutionary Optimization of Neural Networks for Fire Recognition

In this paper, the new hybrid algorithm is used as real time fire recognition algorithm in visual image sequences. For the purposes of real time fire pattern recognition tasks neural networks (NNs) are typically trained with respect of error function minimization by propagating a linear sum of errors. Recent studies in the fire vision recognition have confronted the problem of the inconstant and different shapes of fire which required improving generalization of the NNs. Experimental evidence is presented in this study demonstrating the general application potential of the framework by generating populations of ENN for recognition with a large number of fire shapes in different images, to show that our hybrid algorithm is capable of detecting real time fire vision by improving the generalization of NNs