Forest fires are a critical environmental challenge, causing significant ecological and economic damage each year.Traditional fire detection systems, such as human surveillance and satellite-based monitoring, often face delays and limited accuracy,resulting in slower response times. With the growing availability of image data from satellites and drones, automated detectionsystems have become increasingly viable. This project focuses on using Convolutional Neural Networks (CNN), a deep learningarchitecture, for the rapid and reliable detection of forest fires.CNNs are particularly well-suited for image recognition tasks due to their ability to learn spatial hierarchies of features from rawimage data. The system processes satellite and drone-captured images to detect potential fire outbreaks. By training the CNN on alarge dataset of labeled images, it is able to differentiate between normal environmental conditions and fire-related anomalies. Themodel is designed to enhance both accuracy and speed, addressing the limitations of existing detection methods.The solution reduces false positives and minimizes the risk of undetected fires, ensuring a more proactive approach to firemanagement. Additionally, the model offers scalability, making it adaptable for monitoring vast forested areas in real-time. Inconclusion, this CNN-based forest fire detection system represents a significant advancement in early fire detection, improvingresponse times and potentially reducing the impact of forest fires on the environment