Gaussian Process for the Machine Learning-based Smart Fire Detection System

Artificial intelligence, machine learning, imaging devices, and computer vision are a few examples of contemporary technology that potentially replace traditional methods fire-detection systems with vision-based ones, thereby ensuring society's fire safety. Using the computer-aided system for quick and accurate fire detection could prevent a large-scale fire. Fire detection models perform poorly because they do not take into account the analysis of the fire particles (also known as pattern recognition) or the numerous moving objects in the background. To resolve this issue and concentrate a fire improvement in light of worldly edges from a film, this study gives a powerful even reenactment model of dynamic fire and a half-and-half fire division. In the model's preprocessing step, the candidate fire and non-fire Areas are segmented with the Hue-Saturation-Value (HSV) colour space and the Gaussian's Mixture Model (GMM) to isolate them from the multi-moving item on the complicated background. a unique approach that divides sensory data employing a Gabor filter library with desired orientations and estimates the process of fire. The energy of the wavelengths is also added as an additional aspect to the attribute vector when a temporally-differed image is employed. When all of these traits are added up, a support vector machine (SVM) is produced. to ensure accurate fire detection and better distinguish our data. The suggested fire detection model is validated using multi-conditional video clips. The results of the experiments indicate that the suggested model beats cutting-edge algorithms.