Luu Trong Hieu, Cao Hoang-Long, Ngo Quang Hieu, Nguyen Thanh Tam, Makrini Ilias El, Vanderborght Bram
College of Engineering, Can Tho University, Can Tho 910900, Viet Nam.
Brubotics, Vrije Universiteit Brussel and Flanders Make, Brussels 1050, Belgium.
Data Brief. 2024 Nov 6;57:111118. doi: 10.1016/j.dib.2024.111118. eCollection 2024 Dec.
The popularity of Unmanned Aerial Vehicles (UAVs) in agriculture makes data collection more affordable, facilitating the development of solutions to improve agricultural quality. We present a dataset of rice seedlings extracted from aerial images captured by a UAV under various environmental conditions. We focus on rice seedlings cultivated by the sowing method during their early growth stages because these stages are important to the establishment and survival as well as foundation for lifelong growth. We employed an adaptive thresholding method to isolate rice seedlings from the aerial images. We subsequently classified them into three categories based on their germination conditions: single rice seedings, clustered rice seed plants, and undefined objects. We obtained a total of 5364 labeled images of rice seedlings through data augmentation. This dataset serves as a resource for assessing germination rates and density using machine learning methods. The results derived from these assessments help farmers understand seedling growth and enable them to monitor the health and vigor of rice seedling during early growth stages.
无人机（UAV）在农业中的普及使得数据收集成本更低，推动了提高农业质量的解决方案的发展。我们展示了一个从无人机在各种环境条件下拍摄的航拍图像中提取的水稻秧苗数据集。我们关注通过播种法种植的水稻秧苗在其早期生长阶段的情况，因为这些阶段对其建立和存活以及终身生长的基础都很重要。我们采用了一种自适应阈值方法从航拍图像中分离出水稻秧苗。随后，我们根据它们的发芽情况将其分为三类：单株水稻秧苗、丛生水稻植株和未定义物体。通过数据增强，我们总共获得了5364张带标签的水稻秧苗图像。这个数据集可作为使用机器学习方法评估发芽率和密度的资源。这些评估得出的结果有助于农民了解秧苗生长情况，并使他们能够在早期生长阶段监测水稻秧苗的健康和活力。
