This study presents a computer vision–based method to measure the total feed intake of dairy cows to help farmers manage feeding practices more efficiently. The system uses an overhead camera to capture color images of the feed pile and analyze changes over time. To accurately focus on the feed area, the system uses advanced image processing techniques to eliminate interference from extraneous objects in the image. The total amount of feed is then accurately estimated using computer vision techniques. The method was tested in a real farm environment with an average error of less than 0.35 kg for feed volumes within the range of 0 to 10 kg, outperforming conventional methods. Unlike physical sensors or manual inspection, this noncontact technology ensures that cows behave naturally and remain healthy. By providing real-time, reliable data, it can help farmers optimize feed use, reduce waste, and improve animal welfare.

Read the full article in the Journal of Animal Science.