A noise-robust acoustic method for recognizing foraging activities of grazing cattle

TL;DR

NRFAR introduces a noise-robust online acoustic recognizer for grazing cattle by employing the Chew-Bite Energy Based Algorithm (CBEBA) to classify jaw-movement events into four classes and aggregating them in fixed 5-minute segments to label grazing, rumination, or other activities. Compared with state-of-the-art BUFAR and JMFAR, NRFAR achieves higher frame- and bout-level accuracy and markedly better resistance to environmental noise across matched and mismatched field datasets, DS1 and DS2, with low computational cost suitable for embedded deployment. The method demonstrates strong generalization and resilience to both stationary and nonstationary noises, supporting practical deployment for pasture management and animal welfare monitoring. The work also highlights potential for future improvements, including dynamic segmentation and multi-modal sensor integration for more adverse environments such as barns.

Abstract

Farmers must continuously improve their livestock production systems to remain competitive in the growing dairy market. Precision livestock farming technologies provide individualized monitoring of animals on commercial farms, optimizing livestock production. Continuous acoustic monitoring is a widely accepted sensing technique used to estimate the daily rumination and grazing time budget of free-ranging cattle. However, typical environmental and natural noises on pastures noticeably affect the performance limiting the practical application of current acoustic methods. In this study, we present the operating principle and generalization capability of an acoustic method called Noise-Robust Foraging Activity Recognizer (NRFAR). The proposed method determines foraging activity bouts by analyzing fixed-length segments of identified jaw movement events produced during grazing and rumination. The additive noise robustness of the NRFAR was evaluated for several signal-to-noise ratios using stationary Gaussian white noise and four different nonstationary natural noise sources. In noiseless conditions, NRFAR reached an average balanced accuracy of 86.4%, outperforming two previous acoustic methods by more than 7.5%. Furthermore, NRFAR performed better than previous acoustic methods in 77 of 80 evaluated noisy scenarios (53 cases with p<0.05). NRFAR has been shown to be effective in harsh free-ranging environments and could be used as a reliable solution to improve pasture management and monitor the health and welfare of dairy cows. The instrumentation and computational algorithms presented in this publication are protected by a pending patent application: AR P20220100910. Web demo available at: https://sinc.unl.edu.ar/web-demo/nrfar

Figures (9)

• Figure 1: General block diagram of the BUFAR, JMFAR, and the proposed NRFAR methods divided into temporal scales. The JM information transferred to the top level is different in each method.
• Figure 2: Example of recognized 5-min segments (blue color) compared to the ground truth reference labels (yellow color). The classified activity label assigned to every segment enters the smoothing filter to generate the output label of NRFAR.
• Figure 3: Recording system used to record the acoustic signals composed of microphones (a) that are covered by rubber foam and an elastic headband (b), which are wired and plugged (c) to a digital recorder placed inside a waterproof case (d) attached to a neck halter. Figure extracted from Martinez_Rau2023.10.18.562979
• Figure 4: Top-down scheme for selecting useful audio clips.
• Figure 5: Confusion matrices for different foraging activities for the (a) BUFAR, (b) JMFAR, and (c) NRFAR methods when evaluating on DS1.