Abstract

This work proposes a novel solution to address the challenges of elephant-railway coexistence, such as the risk of collisions, habitat fragmentation, and human-elephant conflicts. The approach involves using a neural network-based monitoring system to track elephant activity near railways. The goal is to create an AI-powered surveillance system capable of real-time detection and tracking of elephants, identifying high-risk areas for collisions, mitigating human-elephant conflicts, and preserving elephant habitats. By integrating cutting-edge technology with conservation efforts, the main objective is to improve railway safety, support sustainable coexistence between elephants and railway infrastructure, and contribute to elephant conservation. The system uses YOLO with a custom dataset to detect elephants with an accuracy of 98.4%, and a custom CNN model for train detection, achieving 91.5% accuracy. With an inference speed of 150ms, the system is suitable for real-time implementation. An Alert signal is sent to the relevant higher authorities, and an infrasonic wave generator can be used to safely drive elephants away from danger zones thereby reducing accidents and conflicts.