S. Chamaille-Jammes

This study evaluates the calibration of deep learning models for camera trap image sequences, showing that temperature scaling improves confidence score interpretability for ecological applications.

The paper reformulates the selection ratio (SR) to provide a unified framework for habitat selection studies, clarifying debates around resource selection functions (RSF) and introducing the selection effect on use (SE) as a complementary metric.

MoveFormer is a Transformer-based model that predicts animal movement steps using past trajectories and environmental covariates, showing the importance of context and history.

This study finds wild boars adjust resting site selection to minimize disturbance risk by prioritizing vegetation cover over proximity to human structures.

This paper introduces dynamic logistic models to analyze time-varying habitat selection without time discretization, demonstrating their effectiveness with simulations and real-world applications on blue wildebeest.

The paper examines how plains zebra survival and reproduction are affected by seasonal changes, finding that reproductive females and their dependent young cope better with longer dry seasons.

This paper evaluates 52 phenology metrics for birth timing in large herbivores, finding that simpler metrics are often better and that circular statistics and distribution tests are most effective.

This study evaluates if citizen science volunteers can reliably detect juvenile animals in camera trap data to study reproduction, finding moderate accuracy for some species.

The paper develops a deep learning pipeline for giraffe re-identification using image similarity networks and CNNs, achieving 90% Top-1 accuracy in identifying known individuals from 4,000 photographs.

This paper develops a simulation model with a genetic algorithm to study how predators and prey use information in different environments, finding that movement unpredictability is key for both, but differs in magnitude and environmental dependence.

This paper introduces a segmentation-clustering method to identify stationary phases in multivariate time-series, applied to animal movement data to detect behavioural modes and home range shifts.