Visual fire detection technologies can detect fire and alarm warnings earlier than conventional fire detectors. This study proposes an effective visual fire detection method that combines the statistical fire color model and sequential pattern mining technology to detect fire in an image. Furthermore, the proposed method also supports real-time fire detection by integrating adaptive background subtraction technologies. Experimental results show that the proposed method can effectively detect fire in test images and videos. The detection accuracy of the proposed hybrid method is better than that of Celik’s method. Visual fire detection technologies can detect fire and alarm warnings earlier than conventional fire detectors. This study proposes enhanced visual fire detection method that combines the statistical fire color model and sequential pattern mining technology to detect fire in an image. supports real-time fire detection by integrating adaptive background subtraction technologies. Experimental results show that the proposed method can effectively detect fire in test images and videos. The detection accuracy of the proposed hybrid method is better than that of Celik’s method.