Early and efficient smoke/fire detection is vital in ensuring the quick arrival and action of the emergency response team. This is to ensure minimal damage is taken by the facility. Therefore, it is important to acknowledge the limits of conventional methods of fire detection such as smoke detector and stationary optic sensors. This includes constraints in providing sufficient data in terms of size and location of fire, the need for smoke particles to reach a certain concentration threshold and sensitivity drop due to the debris accumulation. Hence, the integration of artificial intelligence towards fire detection systems in Unmanned Aerial Vehicles can be used to overcome the flaws of current fire detection systems. Therefore, this paper proposes the Convoluted Neural Network (CNN) model for fire detection and classification with evaluation of processing time, classification accuracy and real time integration to UAV. The study shows that CNN was able to achieve a high prediction accuracy of 93.5. © 2023 IEEE.