Drones for monitoring health of wildlife

The assessment of health and body condition in wild animals has historically posed significant challenges, particularly in regards to automated measurement. First, the motion of animals makes it difficult to obtain 360-degree photographic 'stills' for 3D modelling. Second, distinguishing between the animal and background features (e.g., vegetation) can be challenging.

Recently, unmanned aerial vehicles (UAV), commonly known as drones, have been employed to generate three-dimensional representations of livestock in order to assess their physical traits. This is achieved through the utilization of point-cloud data, obtained through techniques such as 3D photogrammetry (method for creating 3D models from photos) and LiDAR technology (LIght Detection And Ranging or Laser Imaging Detection And Ranging).

Application and fieldwork

In this NLAS sensors project, the researchers use multiple drones, with different sensors, such as RGB cameras and LiDAR, to generate 3D point clouds of wild animals (which was the first challenge). Thermal imaging and ML segmentation models based on Convolutional Neural Nets is used to separate between animals and three-dimensional landscape features, such as dunes and plants (which was the second challenge).

The sensor and software are tested on grey seals. During the breeding and mating season, more than 2000 grey seals gather on the island of Richel between Vlieland and Terschelling. These grey seals are relatively inactive; consequently, multiple photographs can be taken from various angles without the animal moving, thereby overcoming some of the most significant obstacles encountered in previous research.

In this project, the WUR researchers collaborate with Aeria.

Progress (April 2024)

The fieldwork on Richel was a success. A “pipeline” for determine body volume is in development and is progressing steadily. On top of this, the researchers are planning to validate the method at Ecomare on Texel in June.