2018年用于深度学习的带疾病症状注释的玉米田间图像数据集

玉米病害数据集致力于自动化检测北叶枯病(NLB)，一种常见的破坏性玉米叶面真菌病害。该数据集包含以三种不同方式拍摄的玉米叶红绿蓝图像，包括使用手持相机、安装在吊杆上的相机和安装在小型无人机（高度为6m）上的相机。上述图像在已接种NLB型致病病原体的玉米的田间试验中拍摄。生成的数据集包含超过18,222张附带超105,735个注释的NLB 病变图像，或为植物病害方面最大的图像数据集。注释者沿着单个病变点的主轴画线，注视者均为但没有描绘病变点的边缘。在数据集创建者后续的一项研究中，创建者通过继续研究了无人机图像的像素级注释，其中包含非专家根据专家主轴线注释执行病变注释的任务。https://osf.io/p67rz/

This corn disease dataset is dedicated to the automated detection of Northern Leaf Blight (NLB), a prevalent and destructive foliar fungal disease affecting maize. The dataset comprises red-green-blue (RGB) images of corn leaves captured via three distinct acquisition approaches: handheld cameras, cameras mounted on booms, and cameras installed on small unmanned aerial vehicles (UAVs) at an altitude of 6 meters. All images were collected from field trials with maize plants inoculated with the NLB-causing pathogenic agent. The resultant dataset contains over 18,222 NLB lesion images accompanied by more than 105,735 annotations, and is regarded as one of the largest image datasets in the field of plant disease research. Annotators drew lines along the major axis of each individual lesion spot, without delineating the edges of the lesions. In a follow-up study conducted by the dataset's developers, pixel-level annotations for the UAV-captured images were created, where the task required non-experts to perform lesion annotation based on the major axis annotations provided by experts. https://osf.io/p67rz/