Automatic thermal monitoring of calves using low-cost infrared thermography

Infrared thermography offers a non-contact method to record the emitted heat signatures of animals, however its usage is constrained by high costs and requirement for trained staff. To tackle these limitations, we introduce a novel prototype system based on small single board computers and low-cost thermal camera units supported by an automated data management pipeline that allow continuous and automatic monitoring of livestock. We demonstrate the system using calves reared in single-animal pens. The system is composed of sensory platforms and an on-site server, connected wirelessly. The sensory platforms are equipped with a passive infrared motion sensor that triggers the thermal camera when the calf begins to feed or moves to locations relevant for thermal monitoring, i.e. where heat emitting areas such as eyes, ears and muzzle are visible. Thermal images are then collected by the server and stored locally or transmitted over the Internet. However, low-cost thermal cameras are inaccurate, up to ±7 Celsius in farm conditions, according to manufacturer’s data. We explore two methods for performing calibration, validated on the eye temperature of calves. We show that an accuracy of 0.5°C is attainable using on-site calibration. Field tests of the proposed prototype system on a commercial dairy farm showed that it is robust and capable of continuous monitoring without interfering with farm operations.