Deep learning models to identify penguins' prey capture rates

We determined how well deep learning models would work to predict individual prey capture events from tri-axial accelerometer data collected from Chinstrap Penguins. This as done in the Python programming language.Two models were trained (CNN and V-Net architectures) and predictions were made on individual birds that have not been exposed to the model training.The resources include all the source code to produce the figures in the associated publication, as well as example data to run the scripts on.

本研究旨在探讨深度学习模型在预测南极企鹅个体捕食事件方面的有效性，该研究基于从三轴加速度计数据中收集的数据进行分析。该研究采用 Python 编程语言完成。训练了两种模型（卷积神经网络和V-Net架构），并对未接受过模型训练的个体鸟类进行了预测。研究资源包括生成相关出版物中图表的所有源代码，以及用于运行脚本的示例数据。