Precision Livestock Farming Systems Based on Accelerometer Technology and Machine Learning

Livestock production systems play a key role in improving human food demand in terms of quality and quantity. It is essential to monitor well health, reproductive and other issues, activities of cattle and poultry to take timely measures when problems occur. In the animal husbandry industry, cow cattle breeding is an important part and is being invested and thrived with the support of high technology. Cow behavior classification has been reported as a way of early detection of diseases, and interactions in the herd that affect health. Behavior changed when the cow is sick may include decreased daily activities such as eating, drinking, walking, standing, or lying down. Surveillance of cows is centralized in the three most common directions, monitoring and classification of cows’ behavior, injury detection, and determination of the time of reproduction. An operational monitoring system supporting livestock usually consists of a central processor, a computer tasked with receiving information and processing information according to behavior classification algorithms, through information cattle activity. The units gather information about cow activity through sensors and send it to the central processor. In addition, the system can store data and upload data to the internet, which is convenient for monitoring cow’s behavior, managing, and processing information. The main objective of this review is to discuss a methodology to implement precision livestock farming (PLF) sensor systems based on accelerometer technology and machine learning. This methodology uses a feature set and data window to describe leg-mounted or collar-mounted acceleration data to improve the performance of behavior classification.