Collectively advancing deep learning for animal detection in drone imagery: Successes, challenges, and research gaps

Drones have emerged as a powerful tool in animal detection, significantly advancing wildlife monitoring, conservation, and management by capturing high-resolution, real-time imagery over areas often inaccessible or challenging for human observers to reach. However, manual analysis of drone imagery for animal detection is labour-intensive and time-consuming. The application of deep learning methods, particularly convolutional neural networks, in automating animal detection from drone imagery has the potential to revolutionise wildlife monitoring, conservation, and management protocols.

This review provides a comprehensive overview of the increasing use and prospects of deep learning in animal detection using drone imagery. It explores successful applications of deep learning for animal detection, localisation, recognition, and their combinations. The paper also discusses the challenges, limitations, and future research directions of this field. A key message from this review is the need for representative training data covering the various scenarios in which target animals may appear, image annotation difficulties, and the comparability of DL model results across studies. Many studies have focused on single species, locations, or images with a high density of common target species. Assessments of models are potentially biased from using a single test set; many studies lack metrics to evaluate model efficiency, feasibility, and generalizability, and there are uncertainties regarding the optimal number of training images and required ground sample distance (GSD) for different animal detection tasks in drone imagery.

The potential applications of deep learning in wildlife monitoring, conservation, and ecological research using drone imagery are substantial. By enhancing the accuracy and efficiency of animal detection in imagery, this technology could contribute to the understanding and protecting animal populations. To expand the applicability of deep learning to diverse species, environments, and spatial scales, researchers should create standardised benchmark datasets and prioritise open collaboration and data sharing, which would aid in addressing the current challenges.