Automated Detection of Salvin's Albatrosses: Improving Deep Learning Tools for Aerial Wildlife Surveys

TL;DR

The study tackles automated counting of Salvin's albatross on remote Bounty Islands using UAV imagery and a general-purpose BirdDetector, assessing zero-shot versus domain-specific fine-tuning. It combines SAHI-based inference with strengthened data augmentations to improve detection in dense seabird colonies, evaluated through Leave-One-Island-Out Cross-Validation across eight islands. Results show zero-shot performance benefits from SAHI, while fine-tuned models with augmented training significantly improve accuracy, achieving manageable false-positive and miss rates but with island-dependent variability. The work demonstrates the potential for scalable, automated population monitoring in remote environments, with avenues for extending to additional species and counting-centric approaches.

Abstract

Recent advancements in deep learning and aerial imaging have transformed wildlife monitoring, enabling researchers to survey wildlife populations at unprecedented scales. Unmanned Aerial Vehicles (UAVs) provide a cost-effective means of capturing high-resolution imagery, particularly for monitoring densely populated seabird colonies. In this study, we assess the performance of a general-purpose avian detection model, BirdDetector, in estimating the breeding population of Salvin's albatross (Thalassarche salvini) on the Bounty Islands, New Zealand. Using drone-derived imagery, we evaluate the model's effectiveness in both zero-shot and fine-tuned settings, incorporating enhanced inference techniques and stronger augmentation methods. Our findings indicate that while applying the model in a zero-shot setting offers a strong baseline, fine-tuning with annotations from the target domain and stronger image augmentation leads to marked improvements in detection accuracy. These results highlight the potential of leveraging pre-trained deep-learning models for species-specific monitoring in remote and challenging environments.

Figures (2)

• Figure 1: Location of the Bounty Islands relative to New Zealand, and example bounding boxes of the Salvin's Albatross.
• Figure 2: Example predictions for the zero-shot and improved fine-tuned experiments. The coloured boxes correspond to the true positives (green), false positives (red), and false negatives (blue), respectively.