Scholarlink

<p>Plant diseases pose significant threats to agriculture, making proper diagnosis and effective treatment crucial for protecting crop yields. In automatic diagnosis processing, image segmentation helps to identify and localize diseases. Developing robust image segmentation models for detecting plant diseases requires high-quality annotations. Unfortunately, existing datasets rarely include segmentation labels and are typically confined to controlled laboratory settings, which fail to capture the complexity of images taken in the wild. Motivated by these, we established a large-scale segmentation dataset for plant diseases, dubbed PlantSeg. In particular, PlantSeg is distinct from existing datasets in three key aspects: (1) Annotation types: PlantSeg includes detailed and high-quality disease area masks. (2) Image sources: PlantSeg primarily comprises in-the-wild plant disease images rather than laboratory images provided in existing datasets. (3) Scale: PlantSeg contains the largest number of in-the-wild plant disease images, including 7,774 diseased images with corresponding segmentation masks. This dataset provides an ideal yet unified benchmarking platform for developing advanced plant disease segmentation algorithms.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Large-Scale In-the-wild Dataset for Plant Disease Segmentation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月9日

查看全文
  • Tianqi Wei,
  • Zhi Chen,
  • Xin Yu,
  • Scott Chapman,
  • Paul Melloy,
  • Zi Huang

摘要

Plant diseases pose significant threats to agriculture, making proper diagnosis and effective treatment crucial for protecting crop yields. In automatic diagnosis processing, image segmentation helps to identify and localize diseases. Developing robust image segmentation models for detecting plant diseases requires high-quality annotations. Unfortunately, existing datasets rarely include segmentation labels and are typically confined to controlled laboratory settings, which fail to capture the complexity of images taken in the wild. Motivated by these, we established a large-scale segmentation dataset for plant diseases, dubbed PlantSeg. In particular, PlantSeg is distinct from existing datasets in three key aspects: (1) Annotation types: PlantSeg includes detailed and high-quality disease area masks. (2) Image sources: PlantSeg primarily comprises in-the-wild plant disease images rather than laboratory images provided in existing datasets. (3) Scale: PlantSeg contains the largest number of in-the-wild plant disease images, including 7,774 diseased images with corresponding segmentation masks. This dataset provides an ideal yet unified benchmarking platform for developing advanced plant disease segmentation algorithms.