An intelligent fire detection approach through cameras based on computer vision methods

•An intelligent fire detection approach through cameras based on computer vision methods is proposed.•A motion detection method based on background subtraction is used for reducing computations.•An object detection model and an image classification model are constructed and applied for fire detection.•The fire detection rate achieves 98.4% and the false alarm rate achieves 99.9% based on the fire image dataset.•The detection time for one frame achieves 27.4 ms.