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

Tianqi Wei,Zhi Chen,Xin Yu,Scott Chapman,Paul Melloy,Zi Huang

2024-09-06

Abstract:Plant diseases pose significant threats to agriculture. It necessitates proper diagnosis and effective treatment to safeguard crop yields. To automate the diagnosis process, image segmentation is usually adopted for precisely identifying diseased regions, thereby advancing precision agriculture. Developing robust image segmentation models for plant diseases demands high-quality annotations across numerous images. However, existing plant disease datasets typically lack segmentation labels and are often confined to controlled laboratory settings, which do not adequately reflect the complexity of natural environments. Motivated by this fact, we established PlantSeg, a large-scale segmentation dataset for plant diseases. PlantSeg distinguishes itself from existing datasets in three key aspects. (1) Annotation type: Unlike the majority of existing datasets that only contain class labels or bounding boxes, each image in PlantSeg includes detailed and high-quality segmentation masks, associated with plant types and disease names. (2) Image source: Unlike typical datasets that contain images from laboratory settings, PlantSeg primarily comprises in-the-wild plant disease images. This choice enhances the practical applicability, as the trained models can be applied for integrated disease management. (3) Scale: PlantSeg is extensive, featuring 11,400 images with disease segmentation masks and an additional 8,000 healthy plant images categorized by plant type. Extensive technical experiments validate the high quality of PlantSeg's annotations. This dataset not only allows researchers to evaluate their image classification methods but also provides a critical foundation for developing and benchmarking advanced plant disease segmentation algorithms.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper aims to address several key issues in plant disease segmentation:

1. **Dataset Quality and Diversity**: Existing plant disease datasets often lack high-quality segmentation labels and are mostly limited to images taken in laboratory environments, failing to fully reflect the complexity of natural environments. Therefore, the paper proposes a large-scale field plant disease segmentation dataset—PlantSeg.

2. **Segmentation Annotation Types**: Unlike existing datasets that mainly contain class labels or bounding boxes, PlantSeg provides detailed segmentation masks and associates them with plant species and disease names.

3. **Image Sources**: PlantSeg primarily includes plant disease images collected in the field, which enhances the model's applicability in real-world scenarios, as the trained models can be used for integrated disease management.

4. **Dataset Scale**: PlantSeg has a large scale, containing 11,400 images with disease segmentation masks and an additional 8,000 images of healthy plants, categorized by plant species.

With these improvements, PlantSeg not only allows researchers to evaluate their image classification methods but also provides an important foundation for developing and benchmarking advanced plant disease segmentation algorithms.