Classification of digital dermatitis with image processing and machine learning methods

Abstract

In this study, it was aimed to perform the detection and grading of Digital Dermatitis (DD) disease, which is common in dairy cattle and causes serious economic losses, using artificial intelligence techniques in a computer environment with high accuracy without the need for any expert intervention. Within the scope of the study, because of the examinations performed on 168 cows of Holstein breed, aged 4-7 years, whose lameness was detected in dairy farms located in the center and districts of Burdur region, pictures of lesions due to DD were taken, and 4 groups were formed according to the degree of size. The photographs obtained were first labelled according to the degree of disease by a faculty member specialized in podiatry. Afterwards, the tagged photographs were reproduced using artificial intelligence image augmentation techniques, and a sample of 1,000 datasets was carried out for each disease degree. The photographs that make up the dataset were processed using the inception v3 deep learning algorithm and more than 2,000 numerical features were extracted. Then, machine learning models were developed using 6 different machine learning algorithms to classify these features. The results obtained were examined in detail with the help of tables and graphics, and it showed that the developed artificial intelligence models could be used in the classification of DD case photos with a cumulative accuracy value above 0.87.