Elephant-train collisions are a serious problem as a result of the ongoing human-elephant conflict in Sri Lanka. Lack of enough response time as a result of things like poor driver sight at sharp corners, midnight train operations, and severe weather conditions is one of the main causes of these incidents. The majority of these mishaps routinely take place at predetermined spots along known elephant trails and corridors. We suggest creating a unique system using Convolutional Neural Networks (CNNs) for precise elephant recognition and the construction of an early warning system in response to this urgent situation. By utilizing powerful computer vision techniques, this system attempts to improve railway safety in conflict-prone locations. The suggested system will continually monitor its surroundings by installing strategically placed cameras along train tracks. The system will be trained to successfully recognize elephants in realtime video streams using CNNs. When elephants are detected near the tracks, the device activates an early warning mechanism, notifying train operators and allowing them to take preventive actions. By offering a novel way to reduce train-elephant accidents, this study tackles a crucial facet of human-elephant conflict. Modern technology and real-time monitoring combined with the suggested method have the ability to drastically minimize accidents and ensure the safety of both human populations and elephant herds. The adoption of focused mitigation methods may also be facilitated by the capacity to recognize collision prone regions. Through this project, we help to promote peaceful cohabitation between people and elephants while protecting these amazing animals for future generations.