Novel detection of chicken welfare using machine vision

Summary

The Australian chicken meat industry predominantly uses sheds to house chickens. Sheds are typically inspected 3 times per day, however continual monitoring is required for early intervention of chicken stress. In particular, chickens are susceptible to heat stress which is affected by chicken behaviours, e.g. chicken sitting, not grouping together, or grouping together. Previously, USQ-NCEA has demonstrated low-cost remote monitoring using $50 smart cameras, on-board image analysis, to perform tasks like cattle condition scoring, pig weight estimation and crop monitoring in research projects. See project information at www.usq.edu.au/ncea and https://eportfolio.usq.edu.au/user/mccarthc. The project will develop a proof-of-concept system for monitoring chickens and their behaviour in chicken sheds, for the purpose of immediate remote notification of the shed supervisor for events that indicate conditions in the shed that affect flock welfare.

Program

Chicken Meat

Research Organisation

University of Southern Queensland

Objective Summary

1. Develop proof-of-concept low-cost sensing and notification system.
2. Develop image analysis algorithms for detecting chicken behaviour.
3. Make recommendations for further development. A steering committee will be initiated for the project. The monitoring system must be low-cost, as there can be 20 chicken sheds on a farm, and each shed is 100-120 m long by 120 m wide with each shed housing 40,000 chickens. Internet connectivity is not guaranteed at chicken sheds, and the farm office can be over 1 km away from the sheds.