Human-in-the-Loop Eider Duck Counting in Arctic Canada with an Open-Vocabulary Multi-Species Wildlife Detector

Accurate monitoring of eider duck populations in ArcticCanada is essential for understanding ecosystem health andsupporting conservation efforts in a rapidly changing climate.Traditional manual counting from aerial imagery is time-consuming, labor-intensive, and prone to observer bias. Inthis work, we present a human-in-the-loop wildlife countingsystem that integrates an open-vocabulary multi-species ob-ject detector to streamline and enhance the accuracy of ei-der duck surveys. The system leverages a pre-trained open-vocabulary model, enabling the identification of both targetand incidental species without retraining, and employs humanvalidation to correct and refine automated detections. Thiscollaborative workflow combines the scalability of machinelearning with expert ecological knowledge, reducing anno-tation effort while maintaining high accuracy. Field valida-tion using aerial imagery from Arctic Canada demonstratesthat our approach can significantly accelerate population as-sessments, improve consistency across surveys, and facilitateadaptive monitoring in remote environments.