Automated Collection and Analysis of Infrared Thermograms for Measuring Eye and Cheek Temperatures in Calves

SIMPLE SUMMARY: In response to an increasing shift towards automation in the livestock industry, it is important that, as automated systems are developed, they include the capability for automated monitoring of animal health and welfare. As an alternative to current manual methods, this study reports on the development and validation of an algorithm for automated detection and analysis of the eye and cheek regions from thermal infrared images collected automatically from calves. Such algorithms are essential for the integration of infrared thermography (IRT) technology into automated systems to noninvasively monitor calf health and welfare. ABSTRACT: As the reliance upon automated systems in the livestock industry increases, technologies need to be developed which can be incorporated into these systems to monitor animal health and welfare. Infrared thermography (IRT) is one such technology that has been used for monitoring animal health and welfare and, through automation, has the potential to be integrated into automated systems on-farm. This study reports on an automated system for collecting thermal infrared images of calves and on the development and validation of an algorithm capable of automatically detecting and analysing the eye and cheek regions from those images. Thermal infrared images were collected using an infrared camera integrated into an automated calf feeder. Images were analysed automatically using an algorithm developed to determine the maximum eye and cheek (3 × 3-pixel and 9 × 9-pixel areas) temperatures in a given image. Additionally, the algorithm determined the maximum temperature of the entire image (image maximum temperature). In order to validate the algorithm, a subset of 350 images analysed using the algorithm were also analysed manually. Images analysed using the algorithm were highly correlated with manually analysed images for maximum image (R(2) = 1.00), eye (R(2) = 0.99), cheek 3 × 3-pixel (R(2) = 0.85) and cheek 9 × 9-pixel (R(2) = 0.90) temperatures. These findings demonstrate the algorithm to be a suitable method of analysing the eye and cheek regions from thermal infrared images. Validated as a suitable method for automatically detecting and analysing the eye and cheek regions from thermal infrared images, the integration of IRT into automated on-farm systems has the potential to be implemented as an automated method of monitoring calf health and welfare.