Grapevine Leaves RGB Images of Disease Symptons (Esca, Downy Mildew, Powdery Mildew, and Erineum Mite)

This dataset contains 5,267 high‑resolution RGB images of grapevine leaves collected in vineyards located in northern Portugal. Images were acquired under natural daylight using a smartphone without artificial lighting or background manipulation. Leaves were framed to remain fully visible so that symptoms could be documented clearly. The dataset is organised into five classes: - Healthy Leaves, 1122 images from three Portuguese cultivars: Loureiro (494 images), Viosinho (223 images), and Malvasia Fina (405 images). - Downy Mildew (Plasmopara viticola), 1006 images showing yellowish blotches and mold symptoms. - Powdery Mildew (Erysiphe necator), 1126 images with white powdery fungal growth. - Esca Complex, 1000 images presenting tiger-stripe necrosis and vein streaks. - Erineum Mite (Colomerus vitis), 1013 images with blister-like swellings and leaf deformation. Images are provided in JPEG format at two resolutions, 3000 × 3000 pixels and 1024 × 1024 pixels. Healthy filenames include the cultivar name because each plant’s identity was known in those plots and are provided in the format “IMG_YYYYMMDD_HHMMSS_variety”. Symptomatic filenames contain only the timestamp (“YYYYMMDD_HHMMSS”) as many images were collected in mixed‑variety plots where leaf‑level cultivar identification was not reliable. Images come from four vineyards located in the Douro and Vinho Verde regions. These regions have different climate conditions and cultivation practices, which introduce variation in leaf characteristics and disease expression. The dataset can support work in image‑based disease classification, studies of foliar symptoms, cultivar analysis and teaching activities related to plant pathology and viticulture. It can also be used for external testing or combined with other grapevine datasets in larger training workflows. Limitations include the absence of natural occlusion, uneven disease distribution among vineyards and dates, lack of cultivar labels for symptomatic leaves and the absence of segmentation masks or temporal sequences. Annotations are based on visible symptoms without laboratory confirmation.