RT Journal Article
T1 Addressing significant challenges for animal detection in camera trap images: a novel deep learning-based approach.
A1 Mulero-Pázmány, Margarita Cristina
A1 Hurtado-Requena, Sandro José
A1 Barba-González, Cristóbal
A1 Antequera-Gómez, María Luisa
A1 Díaz-Ruiz, Francisco
A1 Real-Giménez, Raimundo
A1 Navas-Delgado, Ismael
A1 Aldana-Montes, José Francisco
K1 Animales - Identificación
K1 Innovaciones tecnológicas
AB Wildlife biologists increasingly use camera traps for monitoring animal populations. However,manually sifting through the collected images is expensive and time-consuming. Current deep learningstudies for camera trap images do not adequately tackle real-world challenges such as imbalancesbetween animal and empty images, distinguishing similar species, and the impact of backgrounds onspecies identification, limiting the models’ applicability in new locations. Here, we present a noveltwo-stage deep learning framework. First, we train a global deep-learning model using all animalspecies in the dataset. Then, an agglomerative clustering algorithm groups animals based on theirappearance. Subsequently, we train a specialized deep-learning expert model for each animal group todetect similar features. This approach leverages Transfer Learning from the MegaDetectorV5 (YOLOv5version) model, already pre-trained on various animal species and ecosystems. Our two-stage deeplearning pipeline uses the global model to redirect images to the appropriate expert models for finalclassification. We validated this strategy using 1.3 million images from 91 camera traps encompassing24 mammal species and used 120,000 images for testing, achieving an F1-Score of 96.2% using expertmodels for final classification. This method surpasses existing deep learning models, demonstratingimproved precision and effectiveness in automated wildlife detection.
PB Springer-Nature
YR 2025
FD 2025-05-09
LK https://hdl.handle.net/10630/38576
UL https://hdl.handle.net/10630/38576
LA eng
NO Mulero-Pázmány, M., Hurtado, S., Barba-González, C. et al. Addressing significant challenges for animal detection in camera trap images: a novel deep learning-based approach. Scientific Reports 15, 16191 (2025). https://doi.org/10.1038/s41598-025-90249-z
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 20 ene 2026