3D posture reconstruction of Siberian Jays in the wild

Recent developments in computer vision and machine learning techniques for animal posture estimation are revolutionizing the data we can collect to study collective behaviour. In particular, the ability to measure fine scaled 3D postures in the wild allows us to quantify the gaze and attention of individuals, which cannot be reliably measured by human observers. In the following project, we hope to develop methods for 3D posture reconstruction of wild Siberian Jays (Perisoreus infaustus) in Sweden Lapland. By combining a multi-camera setup in the wild with posture estimation software with triangulation, we hope to reconstruct the 3D postures of multiple Jays during social learning tasks.

We hope to quantify
1) fine scaled motor behaviour when solving a task, and
2) the attention of spectators by reconstructing the gaze of birds in 3D.

We can then test for how individuals in the wild learn in cooperative groups, the role of kinship in learning and finally fitness consequences. If the field setup is shown to work well, the method can also be extended to exciting research questions and experimental paradigms in the future.