Wildfires pose significant environmental and economic challenges, intensifying due to climate change. This study introduces the Guided Fire Extinguishing Device(GFED), an autonomous air-to-ground system leveraging deep learning for precise wildfire detection, tracking, and response. Using a deep learning model for object detection and the Nona Filter, a priority-based target tracking algorithm, GFED achieves robust performance in adverse conditions such as strong winds, night view, and heavy smoke. Key contributions include optimized deep learning model training, precise mechanical trajectory control, and real-time tracking capabilities. The aerodynamic design, optimized for payload capacity and stability, ensures scalability and reliability. Experimental results demonstrate the effectiveness of GFED, achieving a mean average precision(mAP) of 94.5 % for fire detection. This transformative approach enhances global wildfire response efforts, offering improved safety and efficiency in combating the growing threat of wildfires.