Development and field testing of a system for non-contact estimation of weight in dairy calves using a low-cost 3D camera

Optimal growth and management of replacement animals are crucial for the economic and environmental sustainability of the dairy sector, as it ensures long-term herd size and productivity. Calf body growth is typically monitored using mechanical or electronic scales, however such manual measurements are labor-intensive and can cause stress to young animals. Despite the advances in precision technologies and research focussed on adult cattle, there is a significant gap in solutions specifically designed to estimate the weight and the growth rate of young calves. Hence, this study aimed to develop a system for non-contact evaluation of the calf body weight, using a low-cost 3D camera and an associated algorithm to enable fully automated weight estimate. A total of 230 measurements were carried out on 110 female Italian Holstein calves, aged between 1 to 121 days. Depth (3D) images of each calf’s back were acquired from a top-view configuration, using a Kinect V2 sensor mounted at a height of 1.7 m above the ground. 3D images were analysed using custom-developed software to extract two key morphological parameters for each individual calf, namely the maximum width of the abdomen (WA) and the height (HM) measured at the same position,. Simultaneously, each imaged calf was weighed on a mechanical scale with a resolution of ± 1 kg. In addition, its abdomen width and height at the withers were manually measured with a resolution of ± 5 mm, for further validation of the system accuracy. The two image-extracted morphological parameters were finally used to train a simple prediction model for calf body weight. The resulting equation demonstrated a fairly good prediction capability (R2 = 0.87, RMSE = 3.52 kg, mean relative error = 6.6 %) when applied to a validation set of calves weighing between 24 and 88 kg. These results, combined with the simplicity of the measurement setup and the low cost of the device used, highlight the potential of implementing daily monitoring of individual calf growth using such a system, with significant advantages over traditional, labor-intensive weighing method based on scales.