IoT-Based Real-Time Wild Animal Detection and Alert System Using Deep Learning and Multi-Sensor Integration

Abstract

Throughout the world - especially those areas close to Forest Ecosystems - humans often face difficulties or challenges created as a result of conflict with various types of animals around them. Such conflicts arise  when livestock have been attacked by wild animals, or when there is a direct threat to the safety of humans from an attacking wild animal(s) that may cause harm or even death. This article introduces a new IoT based intelligent wildlife monitor/detection system. It will use the YOLOv8 deep learning algorithm/architecture combined with Environmental Sensor Networks (ESNs), as well as ESP32 microcontrollers for Real-time Data Acquisition (RDA) from 3 types of sensors: DHT11 Temperature & Humidity Sensor, MQ135 Air Quality Sensor and HC-SR04 Ultrasonic Distance Sensor; and provide the user with a way to monitor wildlife movement through actual real-time webcam images using access control (RBAC) based website (Flask Framework), receive alerts about detected wildlife on multiple channels (Telegram API, Google Text-to-Speech audio notifications and LCD displays) and maintain historical records of previous detection events and all user accounts through an SQLite Database. By testing the prototype and analyzing its findings against the traditional methods used; it offers a much better response time to classify detected animals (under 2 seconds) and provides a much longer period of time that the user has to respond than traditional methods of monitoring/watching for wildlife attacks.