Cow udder detection in thermal images of cows

Abstract

Mastitis (inflammation of the udder) causes agricultural businesses significant losses. It is likely that mastitis detection methods based on thermal imaging would be faster and encourage more testing. However, they automated detection of udder, which would be the first step towards using those images, hasn’t received much attention. Thus this paper discusses a possible way to detect udder in thermal images. The proposed method is based on marker guided watershed transform. It has achieved the average sensitivity of 0.9320, average specificity of 0.8189 and average MCC of 0.6159.