Project-AgML/gvlid_grapevine_leaf_disease

Gvlid Grapevine Leaf Disease

GVLiD (GrapeVine Leaf identification of the Diseases) is a grapevine leaf disease classification set of 3,477 field photographs at 1080 x 1080 pixels in four classes as published: healthy (1,109), black rot (808), Esca (888) and leaf blight (672). The photos were taken from different angles during vineyard visits in Maharashtra, India, with smartphones, by authors at Annasaheb Dange College of Engineering and Technology. The record's metadata.json gives each photo a vineyard (28 named vineyards with GPS points east of Pune), a capture time, lighting, weather and a device; these are kept as columns as published, with the caveat that they look templated (every photo on 2023-06-15 with one device, and lighting and weather each split evenly three ways). The disease_type_metadata column repeats the metadata's disease field, which disagrees with the class folder for one photo. Mendeley Data version 5. The dataset contains 3,477 images across 4 classes: black_rot, esca, healthy, leaf_blight.
Images per class:

• black_rot: 808
• esca: 888
• healthy: 1,109
• leaf_blight: 672

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

Citation

<!-- TODO: add BibTeX citation -->

The dataset itself can be cited as:

Shikalgar, A., Savalkar, A., Bhasme, A., Chavan, S., & Nikam, V. (2026). GVLiD: GrapeVine Leaf identification of the Diseases [Dataset]. Mendeley Data. https://doi.org/10.17632/WKYMF8BHCG.5

This dataset was reformatted from its original format to match HuggingFace standards.