Human-in-the-loop visual wildlife monitoring systems

Abstract

This project focuses on developing practical tools for wildlife monitoring, specifically addressing two key conservation needs: counting large bird flocks and analyzing camera trap images from wildlife habitats. Our work centers on two platforms: BirdCount, designed for monitoring Amur Falcon populations in Northeast India, and CaTRAT (Camera Trap Recognition and Analysis Tool), which processes and analyzes camera trap images for wildlife identification and population studies. The first phase of development establishes the foundation through a modern web interface built with React, focusing on user-friendly image annotation tools and robust data management systems. This infrastructure supports essential features like image upload, automated counting, manual correction capabilities, and administrative oversight for data validation. The second phase will enhance the systems’ analytical capabilities, particularly in species identification and population counting. This project aims to provide reliable tools that can be effectively used in field conditions by wildlife researchers and forest department personnel.