Abstract

This article describes a system for automatic monitoring of cattle behavior using video cameras in grazing fields (80 [has]). This system is made up of an IoT cow's collar in charge of locally store signals obtained from inertial movements units and microphones. The IoT cow's collar also is comprised of GPS and a LoRa transceiver to send in real-time the location of the cow to a central point. The central point consists of a 9 meters long post with two PTZ-IP video cameras at the top. The cameras, with enhanced zoom and night vision capabilities, are connected to a DVR to store the video streaming. The video cameras are programmed to continuously track the animal using the GPS signals from the cows. This paper describes the use of polynomial fitting to ensure that video streaming keeps the target cow in the center of the video frame. This automatic framing procedure produces good enough videos to identify different types of movements only from the stored video streaming.