Methane emissions from livestock are a major environmental concern, contributing approximately 14% of total anthropogenic greenhouse gas (GHG) emissions from agriculture. Among ruminants, eructation (belching) is a key physiological process through which most methane (CH \(_4\) ) is released. In this study, we present a livestock-wearable system designed to detect belching events using inertial information, incorporating 6-degree-of-freedom inertial measurement units (IMU) placed around the animal’s head to capture mechanical vibrations linked to eructation. A low-power commercial micro-electromechanical methane gas sensor was integrated in order to simplify the annotation process during the data collection stages, enabling a rapid, scalable and human-independent labeling strategy. Machine learning (ML) models were evaluated and trained to anticipate eructation events based only on the IMU sensor data, while using the commercial CH \(_4\) sensor to label significant events (emissions above a certain concentration threshold) in real-time. Beyond this labeling process, during the training and testing stages, the CH \(_4\) sensor information is not required and all estimations rely only on the IMU inertial readings. Field validation demonstrated prediction accuracies of up to 79.7% for individual subjects, providing results that suggest substantial potential for the accurate estimation of these belching events, under natural grazing conditions. These findings highlight the potential of integrating IMU-based sensing and ML algorithms as a scalable, minimal invasive alternative for methane monitoring in livestock. The approach can support better understanding of methane emission dynamics and inform mitigation strategies in precision livestock farming.
