Autonomous Monitoring Systems for Detecting Illegal Wildlife Activities and Habitat Encroachment

Abstract

The poaching of wildlife, encroachment on natural habitats and late surveillance remain a challenge for the protection of biodiversity and protected ecosystems around the globe. Traditional monitoring approaches rely on manual patrol and on individual sensors, leaving areas uncovered, a delay in detection and high cost of operation. This review will comprehensively analyse the technology, benchmark data, AI models and deployment of autonomous systems designed to detect illegal wildlife activities and habitat encroachment. The camera trap data from Snapshot Serenet is regarded as an excellent example of publicly released wildlife monitoring data for use as a benchmark for research. The proposed framework is monitored using camera traps, UAVs, acoustic sensors, IoT devices, edge computing, and YOLOv8 deep learning model for real-time detection and intelligent event recognition. By comparing with the results of traditional CNN, SSD, Faster R-CNN, and YOLOv5 based methods, it can be seen that the accuracy of detection, precision, recall, F1 score and ROC-AUC rate of autonomous AI-enabled systems are all higher and the false alarm rate is lower, while response efficiency is higher. This review's novelty is that it brings together the three key concepts of multimodal sensing, edge intelligence, and explainable artificial intelligence (XAI) to facilitate autonomous conservation monitoring. The paper provides a thorough classification of monitoring technologies, comparison of their performances, a summary of the existing research gaps, and a set of future research directions for scalable, privacy-aware and intelligence-based wildlife protection systems. In general, autonomous monitoring systems offer effective and timely decision-making support to reinforce the conservation and sustainable use of habitat for wildlife.