PlantVillage Apple Synthetic Segmentation Dataset

# Restricted PlantVillage Subset with Segmentation Masks This dataset is a **manually restricted subset of the [PlantVillage dataset](https://data.mendeley.com/datasets/tywbtsjrjv/1)** (Hughes & Salathé, 2015). A balanced subset of apple leaf images was selected across three disease classes: Apple Scab, Apple Black Rot, and Cedar Rust. Images were manually selected to ensure varying disease severity while avoiding structural bias, creating a controlled low-resource setting for downstream evaluation. The original PlantVillage dataset does not include pixel-level annotations. For this subset, **pixel-accurate segmentation masks of diseased regions were manually created** using [CVAT](https://www.cvat.ai/) (Computer Vision Annotation Tool) and are provided alongside each image. ## Data Split - **Train:** 30 images (10 per disease) - **Validation:** 15 images (5 per disease) - **Test:** 30 images (10 per disease) **Total:** 75 images ## Naming Convention Files follow the naming convention `<disease>_i`, where *i* ranges from 1 to *n* (the number of images in that split). This convention is consistent across both images and their corresponding segmentation masks. ## Reference Hughes, D. P., & Salathé, M. (2015). *An open access repository of images on plant health to enable the development of mobile disease diagnostics*. arXiv:1511.08060 https://arxiv.org/abs/1511.08060