Livestock Images and Behavior (2024-2025)

Today, many research articles have demonstrated the possibility of image analysis for monitoring livestock behaviour: posture estimation, location, inter-individual distances … Many information that can be used to better understand animal welfare and behavior, and that can help improving livestock management.

So why is that tool not used at a large scale by researchers? First, it can be technical to deploy IA algorithms, even if there were great improvements the last couple of years. But more important, it generally needs to built a dataset of annotated images, train a neural network, test its efficacy … and repeat until obtaining satisfying results.

This last stepq can be overcome by building a high quality, large, and highly diverse dataset, with images coming from many environments.

This is the objective of LIB: sharing and collecting images from international projects to build a dataset for enhancing the research of tomorrow.

Not all behavior could be annotated and LIB will concentrate to the most common information, that are a prerequisite of many studies: detection of animals, animals’ head and posture.

The results of LIB, part of the TEF Agrifood European project, will be a data paper, and the dataset accessible from a public repository.

Thank you to all the contributors, what are you waiting for to contact us and share!?

Some examples of annotated images and more info below !

Goats grazing outdoor @mathieu.bonneau

Lactating sow @Laurianne.canario

Calfs @Lucile.Riaboff

Lactating goats @Masoomeh.Taghipoor

Sheep in behavior test @Dominique.Hazard and David.Barriere