Pearl millet, a well-known dryland climate-resilient cereal predominantly cultivated nearly 30 million hectares especially in high temperature zones of Asia and Africa. Both the continents share a major contribution of global millet production. Though the crop can withstand high temperature and water scarcity, the yield potential is constrained by insufficient monitoring and changes in climatological factors ultimately leading to the development of serious diseases like downy mildew, rust, ergot, smut, and blast. Conventional disease detection methods are labour intensive and often require professional expertise, finding difficulty for the farmers and reducing the yield. Early detection is important for effective management of these diseases at an appropriate time. With intervention of computer technology, a deep learning (DL) system helps in real-time detection of leaf diseases under field condition. In this paper, the YOLO algorithms were evaluated using 2473 field images collected from different parts of the Coimbatore district representing all diseases of pearl millet including healthy leaves. The comparisons of YOLO v5, v8, and v11 with state-of-the-art models revealed that the YOLOv11 algorithm was found to be the best with an accuracy of 96.73% over other models.