Summary

A recently completed project PRJ-010646 ‘Novel detection of chicken welfare using machine vision’, funded by the AgriFutures Chicken Meat program in 2017-2021, developed and evaluated proof-of-concept machine vision algorithms for detecting flock weight and behaviour in commercial shed conditions. The technology used low-cost off-the-shelf video cameras and achieved weight estimation within 5% of manual flock weight measurements, and automatically discriminated between moving and stationary birds and behaviours including respiratory rate, an important indicator of heat stress in chickens. Additionally, the project identified image analysis-based flock behavioural attributes that were highly correlated with Dawkins’ optical flow (Dawkins et al., 2009, 2013) which is an early predictor of footpad dermatitis and hockburn.

 

The University of Southern Queensland and AgriFutures are seeking commercial partner(s) to develop and commercialise the technology for use by rural industries, and discussions have commenced with a potential commercial partner. A project is required to facilitate engagement with the commercial partner, perform porting of algorithms onto commercial camera hardware, conduct commercial scale farm trials and develop a software user interface for flock alerts.

 

The project addresses Objective 3 of the RD&E plan, ‘Improve chicken meat production through the whole supply chain’ in the key focus areas of ‘Enhancing chicken welfare and handling methods’ and ‘Invest in more automation (initiatives for innovative ways to improve efficiencies)’.

Program

Chicken Meat

Research Organisation

University of Southern Queensland