A professor with the University of Saskatchewan reports progress in the development of a new tool to assess indicators of on farm animal welfare at slaughter.
Researchers with the University of Saskatchewan's Western College of Veterinary Medicine and Department of Electrical and Computer Engineering with funding provided through the NSERC Industrial Research Chair in Swine Welfare research program have developed an automated swine welfare assessment system using computer based artificial intelligence to evaluate carcasses at the abattoir.
The system consists of a digital security camera and an artificial intelligence processor.Dr. Seokbum Ko, a professor in the Department of Electrical and Computer Engineering, says throughout processing this system identifies and evaluates lesions, serving as on farm welfare indicators.
Quote-Dr. Seokbum Ko-University of Saskatchewan:
Initial findings suggest this approach is very cost effective and provides insights into on farm welfare and preslaughter handling.Utilizing computer vision ensures consistent unbiased data collection crucial for monitoring and improving animal welfare standards.
Regarding the technology's capability to evaluate lesions, the AI models are displaying improved accuracy in identifying and categorizing lesions on pig carcasses.Several factors impact this ability.Firstly, the quality and quantity of data utilized for training significantly influences performance.A diverse dataset with ample images representing various lesion types and severity is crucial for effective training.
Secondly the complexity and appearance of lesions can pose challenges for accurate detection, especially subtle or irregularly shaped ones.Moreover, environmental factors like lighting conditions in the abattoir can affect image quality and consequently lesion detection accuracy.Optimising camera placement and lighting is essential to ensure clear and consistent images for analysis.
Dr. Ko says the promising advancements validate the potential scalability of this technology for widespread commercial implementation.He says the next step is to refine the computer models to ensure they deliver more precise results.
Source : Farmscape.ca