Swine behavior monitoring is pivotal for precision livestock farming, and posture recognition remains challenging. We propose a hybrid LSTM-CNN algorithm utilizing neck-mounted inertial sensors to classify four typical postures standing, feeding, lying, and active movement. By extracting time-domain features such as signal magnitude and variance from triaxial acceleration data, the architecture integrates CNN’s spatial pattern learning with LSTM’s temporal sequence modeling. The model achieves 73.2% recognition accuracy on 2,276 annotated samples, surpassing traditional machine learning baselines by 10–15% across the entire dataset. Our sensor fusion approach demonstrates practical value for real-time monitoring in commercial farms, offering critical behavioral biomarkers for health management. The system is deployable on edge devices with high computational efficiency, advancing intelligent livestock welfare assessment.