DeepWeeds
收藏资源简介:
DeepWeeds数据集是由澳大利亚詹姆斯库克大学科学与工程学院创建,包含17,509张澳大利亚北部草原地区八种重要杂草的标记图像。该数据集旨在支持深度学习模型在复杂自然环境中对杂草种类的鲁棒分类,以促进农业机器人杂草控制技术的发展。数据集涵盖了光照、旋转、尺度、焦点、遮挡、动态背景以及地理和季节变化等多因素变量,适用于训练复杂深度学习模型。DeepWeeds数据集的应用领域主要集中在提高草原杂草管理的自动化水平,解决杂草对农业生产的影响问题。
The DeepWeeds dataset was developed by the School of Science and Engineering at James Cook University, Australia. It comprises 17,509 labeled images of eight major weed species in the grassland regions of northern Australia. This dataset is designed to support robust species classification of weeds by deep learning models in complex natural environments, thereby advancing the development of agricultural robotic weed control technologies. The dataset incorporates a wide range of confounding variables including illumination conditions, rotation, scale, focus, occlusion, dynamic backgrounds, as well as geographical and seasonal variations, rendering it well-suited for training advanced deep learning models. The primary application scenarios of the DeepWeeds dataset center on enhancing the automation of grassland weed management and mitigating the adverse impacts of weeds on agricultural production.




