Evaluation of video-based determination of the respiratory rate in calves

This work aims to evaluate whether it is possible to assess the respiratory rate of calves retrospectively using camera recordings. Respiratory rate is a parameter for detecting heat stress. Heat stress in livestock is an increasing problem due to global warming, affecting both animal welfare and the profitability of farms. By default, respiratory rate is measured by direct observation by staff. This method is both time-consuming and labour-intensive, does not allow continuous monitoring and can additionally stress the animals. As part of the experiment, ten Simmental calves aged three to ten weeks were filmed with surveillance cameras installed in the calf barn. Respiratory rates were determined both by direct observation by two observers and retrospectively from the video recordings by the same observers. A total of 330 respiratory rate sequences were analysed for agreement between live observations and video-based counts. A statistical analysis was performed, including Spearman correlation and Bland-Altman plots, to evaluate the accuracy and reliability of the video-based method. The results indicated a significant correlation between live counts and video recordings, with video-based counts tending to underestimate respiratory rates, especially at higher levels. The study concludes that video recordings are generally suitable for monitoring the respiratory rate of calves, but that improvements in image quality, camera positioning, and lighting are needed to ensure reliable detection. In the future, artificial intelligence (AI) could be used for automated monitoring to detect heat stress in calves more efficiently and improve animal welfare. Studies, like the present one, are a first step in developing such AI-based approaches.