2019年DeepSeelding项目中使用的数据集

DeepSeedling数据集为使用深度学习模型在田间检测和棉苗计数创建。原始数据以视频形式使用三台分辨率相同的1920×1080像素彩色相机拍摄，并在两个生长季节分别在三个田间位置进行数据收集。图像仅涉及种植后7-11天的早期生长阶段幼苗。从视频剪辑中提取 红绿蓝 图像以构建用于植物播种检测的数据集。在数据集中，在不同位置获取的三组图像，分别包含2391、1821和1531张jpg 格式图像，以及对应的棉苗和一小部分杂草的边界框注释。https://figshare.com/s/616956f8633c17ceae9b; https://github.com/UGA-BSAIL/deepseedling

The DeepSeedling dataset was created for in-field detection and cotton seedling counting using deep learning models. The raw data was captured as videos using three identical 1920×1080 pixel color cameras, and data collection was conducted at three field locations across two growing seasons. All images only cover seedlings at the early growth stage, 7–11 days after planting. RGB images were extracted from video clips to construct this dataset for plant stand detection. The dataset contains three groups of images acquired from different locations, with 2391, 1821, and 1531 JPG-format images respectively, along with corresponding bounding box annotations for cotton seedlings and a small portion of weeds. The dataset is available at: https://figshare.com/s/616956f8633c17ceae9b; https://github.com/UGA-BSAIL/deepseedling