Article Dans Une Revue Peer Community In Ecology Année : 2025

Democratising AI for Ecology: MEWC Delivers Custom Wildlife Classification to All

Résumé

Artificial intelligence (AI) has emerged as a transformative solution for processing vast amounts of camera-trap imagery, yet its integration into ecological research has been hindered by technical complexity and limited accessibility. Many existing tools are either difficult to customise, rely on costly infrastructure, or require advanced coding expertise, excluding a significant portion of conservation practitioners and citizen scientists. In response to these challenges, a new AI pipeline—MEWC (Mega-Efficient Wildlife Classifier)—offers a robust, open-source alternative tailored for ease of use and adaptability. MEWC integrates deep learning technologies with modular, cross-platform compatibility, enabling users to train and deploy custom wildlife classifiers with minimal technical background. Demonstrating high accuracy and scalability, particularly in a Tasmanian wildlife case study, the system also supports integration with community tools, promoting seamless workflows for species identification and data validation. Importantly, MEWC preserves user control over data and addresses growing concerns about privacy and data sovereignty. While limitations remain—such as dependency on high-quality training data and potential regional constraints—the pipeline's design fosters transparency, flexibility, and future extensions into behavioural and cognitive ecology. MEWC represents not just a technical advancement but a paradigm shift toward democratising AI-powered ecological monitoring.

Fichier principal
Vignette du fichier
SUEUR 2025 PCI Ecology.pdf (981.51 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05071652 , version 1 (17-05-2025)

Licence

Identifiants

Citer

Cédric Sueur. Democratising AI for Ecology: MEWC Delivers Custom Wildlife Classification to All. Peer Community In Ecology, 2025, pp.100787. ⟨10.24072/pci.ecology.100787⟩. ⟨hal-05071652⟩
148 Consultations
135 Téléchargements

Altmetric

Partager

  • More