Authors: Patole Omkar Jagdish, Prof. M. R. Kumbhoje, Prof. M. C. Kolse, Prof. S. K. Said

Abstract: Automated wildlife monitoring is central to biodiversity conservation, protection of endangered species, habitat preservation and reduction of human–wildlife conflict. Conventional approaches that rely on manual inspection of camera-trap images and field acoustic recordings create large processing backlogs, delay responses to threats and suffer false positives from moving vegetation. This paper reviews AI for wildlife monitoring. It examines deep convolutional networks for detecting, classifying and counting animals in camera-trap images, generalization across new locations, detectors that filter empty images, individual re identification of striped and spotted animals such as tigers and leopards using deep embeddings, bioacoustic monitoring of birds and other species, and detection of threats such as gunshots and chainsaws. Anti-poaching decision support and drones with thermal detection are discussed, along with deployment on solar-powered edge devices with compressed models. A multi-tier architecture comprising field sensing, edge inference, connectivity, a conservation intelligence platform and ranger response is proposed, together with research gaps, objectives, a comparison of methods and challenges such as domain shift, rare species, power and connectivity in forests. The review concludes that AI can turn monitoring data into timely conservation action.

DOI: https://doi.org/10.5281/zenodo.24219160