
The Zambia Maize Leaf Dataset is a collection of carefully selected and annotated photos created for the purpose of detecting maize leaf disease in the field. The dataset, which is divided into four classes, Maize Streak Virus (MSV), Gray Leaf Spot, Northern Leaf Blight, and Healthy Leaves, consisting of 19,990 RGB photos that were gathered from Zambian maize fields. To enable direct use in YOLO-based training pipelines, a relevant YAML configuration file is supplied. All photos are tagged using bounding boxes in accordance with the YOLO annotation protocol. The dataset is appropriate for assessing model resilience in actual agricultural settings because it reflects natural fluctuations in illumination, background clutter, and disease severity.