With the growing demand for forest fire prevention, intelligent smoke and fire monitoring and early warning technology has become a key measure to prevent forest fires. Traditional monitoring systems often suffer from poor image quality, low recognition accuracy, and limited positioning and communication capabilities. To overcome these challenges, this project has innovatively developed an intelligent smoke and fire monitoring and early warning system. The system focuses on three core technologies: image de-hazing and enhancement, intelligent smoke and fire recognition, and precise positioning and communication. In terms of image processing, an adaptive de-hazing and enhancement model based on an improved CycleGAN network has been developed, effectively addressing image degradation under complex weather conditions. For target detection, an improved YOLO algorithm with multi-scale feature fusion has been applied to enhance the accuracy of smoke and fire identification. In terms of alarm communication, the system innovatively integrates BeiDou short message technology, enabling positioning and transmission of fire alerts in areas without network coverage. This system provides an all-weather, highly reliable intelligent monitoring solution for forest fire prevention, significantly enhancing China's capabilities in forest fire prevention and emergency response.