Využití metod umělé inteligence pro detekci požáru z obrazových dat

Abstract

This thesis focuses on the design and implementation of an intelligent model for fire detection from image data using artificial intelligence methods. The aim is to develop a system capable of autonomously distinguishing between scenes with and without fire indicators and to evaluate its functionality under various conditions. The thesis also involves the creation of synthetic image data through simulation techniques and generative artificial intelligence methods, which expand the training dataset and enable testing of the model in diverse scenarios. The theoretical part summarizes the issue of fire detection, image data processing methods, and the principles of machine learning and deep learning. The practical part deals with dataset construction, the design and training of the model, and the subsequent analysis of its parameters and behaviour. The thesis aims to assess the suitability of the proposed solution for use in early fire detection systems.