V-Net for Animal Behavior

Adapting the V-Net fully convolutional neural network from biomedical image segmentation to temporal sensor data for animal behavior classification.

The V-Net is a fully convolutional neural network originally designed for 3D biomedical image segmentation. We adapted this architecture for 1D temporal data from animal-borne sensors, creating a powerful tool for behavior classification that can process continuous streams of accelerometer and gyroscope data.

Our adapted V-Net achieves an AUC score of 88% for predicting six behavioral categories in green turtles, and has since been successfully applied to chinstrap penguin prey capture detection and other species.

Key contributions

  • Novel adaptation of V-Net from 3D image to 1D temporal signal processing
  • 88% AUC for six-class behavior prediction in green turtles
  • Application to chinstrap penguin prey capture quantification
  • Open-source implementation for the ecology community

References