Effective control of animal feed and pasture management are increasingly important factors for animal health and farm sustainability. Recent technological advances in animal monitoring devices offer a significant potential for enhancing these practices. This paper presents the development of an innovative animal monitoring system for sheep, designed to capture images of pastures while minimizing redundant data collection. By integrating Machine Learning (ML)-based animal posture detection, the system autonomously triggers image acquisition only during relevant feeding activities. Additionally, the system automatically uploads the captured images for processing, reducing the need for manual intervention. Preliminary results demonstrate the system's feasibility and improved efficiency compared to state-of-the-art approaches.