Social networks in pigs based on computer vision: impacts of proximity definitions

C.A.E.M. Orsini*, B. Hegedüs, L.E. van der Zande, I. Reimert, P. Bijma, J.E. Bolhuis

*Corresponding author for this work

Research output: Contribution to conferenceAbstract

Abstract

In farm animals, social network analysis has become a popular approach to study agonistic interactions, behaviour disorders, or preferential associations. Previous studies focused on a single type of social interaction or proximity definition. Recent advances in technology enable comparisons of networks built with various proximity definitions. The aim of this study was to explore how different spatial proximity definitions affect networks based on computer vision. For this purpose, video data were collected over three consecutive days in 21 pens, each containing six pigs. A tracking-by-detection method was applied to detect the location and track each individual by using bounding boxes. Networks were constructed with five different definitions of proximity, four categorical definitions: (1) distance between centers of bounding boxes <50cm, (2) occurrence of overlap of surfaces of bounding boxes, (3) surface of overlap of bounding boxes >20%, (4) a combination of (1) and (3), and one quantitative definition: (5) the harmonic mean of the distance between the two individuals. The agreement between (2) and (4) was the lowest at 81.9%, whereas the agreement between (1) and (4) was the highest, reaching 98.9%. Our results reveal correlations between centrality metrics across social networks built with different proximity definitions. Direct metrics, such as degree centrality, were less affected by proximity definitions (with correlations ranging from 0.66 to 0.99) compared to indirect metrics, like eigenvector centrality (ranging from 0.26 to 0.86). The results highlight the need to exercise caution in the choices made when assessing or comparing social networks based on proximity.
Original languageEnglish
Publication statusPublished - 24 Oct 2023
EventADP Science Day 2023 - Landgoed Welderen, Elst, Netherlands
Duration: 24 Oct 202324 Oct 2023

Other

OtherADP Science Day 2023
Country/TerritoryNetherlands
Period24/10/2324/10/23

Fingerprint

Dive into the research topics of 'Social networks in pigs based on computer vision: impacts of proximity definitions'. Together they form a unique fingerprint.

Cite this