Stage M2 : IA-based automated detection and behavior analysis among piglets (Rennes)

5 December 2023

---

Catégorie : Stagiaire

---

 

Keywords : Animal welfare, Artificial Intelligence, Computer Vision, Automated Behavior Analysis, Action Recognition, Animal Pose Estimation, CNN, Transformers, LSTM

 

 

Context

The proposed internship is part of a project on animal welfare, which goal is to use Artificial Intelligence and new technologies for tracking key indicator traits in animals facing challenges of the agro-ecological transition. Animal welfare is a key agro-ecological process to be optimized for livestock production. Automated behavior analysis (ABA, also called computational ethology) is the use of technology to detect and observe the behavior of animals in ways that require minimal human labor.Play behaviour is considered an indicator of animal welfare in young pigs. However, as play behaviour events are short-lasting and occur sporadically, continuous monitoring is necessary. We could use standard video cameras, available in most modern farms, to monitor livestock. However, most computer vision algorithms perform poorly on this task, primarily because, (i) animals bred in farms look identical, lacking any obvious spatial signature, (ii) none of the existing trackers are robust for long duration, and (iii) real-world conditions such as changing illumination, frequent occlusion, varying camera angles, and sizes of the animals make it hard for models to generalize.

 

Mission

Given a set of annotated video recordings of piglets provided by INRAE. the objective is to measure the social interactions and the occurrence of fights vs positive social contacts between piglets before and after weaning to assess emotional states of piglets. There are currently several methods for performing pose detection for animals, learned from different corpora, as well as for performing action recognition, using for example CNNs feature coupled with LSTMs. The goal of this internship is to evaluate some of these methods on the provided corpus. This will require implementing methods and adapting them (possibly through transfer learning).

 

Activities

• ·Develop or adapt sota algorithms for group-housed pigs to perform :
  • oanimal pose estimation
  • oinstance level segmentation,
  • otracking
  • oaction recognition
  • ·Evaluate the performances
  • ·Results analysis to pinpoint difficulties of both the dataset and the methods

Skills

• ·Master in Computer Sciences, with proficiency in python and its libraries for deep learning
• ·General background in computer vision and machine learning
• ·Understanding of deep learning methodologies and techniques
• ·Proficiency in data handling, particularly in video processing