基于改进型YOLOv7算法的玉米雄穗识别检测OA
Identification and Detection of Maize Tassels Based on Improved YOLOv7 Algorithms
监测玉米抽雄期生长状态在其最终产量中起重要作用,使用无人机搭载可见光相机采集玉米抽穗期RGB图像,研究目标检测算法,使用合适的模型自动识别图像中的玉米雄穗.针对原始数据,将图片裁剪为640×640的大小,使用裁剪后的图像构建目标检测模型训练和测试所需数据集.对YO;Ov7模型进行改进,通过增加一个检测头和引入SIoU损失函数这2种改进策略,增强模型性能.改进后模型在测试数据集中全类平均正确率(mAP)值可达到97.2％,相较改进前提升了 2.7％.实验结果表明,所提出的改进策略能有效提高YOLOv7模型对遥感图像中玉米雄穗的检测准确率,可以较准确地完成玉米雄穗的识别与计数任务.
Monitoring the growth status of maize during the tasseling period plays an important role in its final yield.A drone equipped with a visible light camera is used to capture RGB images of maize during the tasseling period.Its object detection al-gorithms are studied,and appropriate models are used to automatically identify maize tassels in the images.The images are cropped to obtain a size of 640×640 picture based on the original data,and the cropped images are used to construct the re-quired dataset for training and testing the object detection model.The YOLOv7 model is improved by adding a detection head and introducing SIoU loss function which are two improvement strategies for enhancing model performance.The improved model achieves an mean average precision(mAP)value of 97.2％in the test dataset,an increase of 2.7％compared to before the improvement.The experimental results show that the proposed improvement strategy can effectively improve the accuracy of YOLOv7 model in detecting maize tassels in remote sensing images,and can accurately complete the task of identifying and counting tassels.
古明琦;班松涛;李琳一;胡冬;田明璐
上海第二工业大学,计算机与信息工程学院,上海 201209||上海市农业科学院,农业科技信息研究所,上海 201403上海市农业科学院,农业科技信息研究所,上海 201403上海市农业科学院,农业科技信息研究所,上海 201403上海市农业科学院,农业科技信息研究所,上海 201403上海市农业科学院,农业科技信息研究所,上海 201403
信息技术与安全科学
深度学习目标检测玉米雄穗识别YOLOv7
deep learningobject detectionmaize tassel identificationYOLOv7
《微型电脑应用》 2026 (3)
40-43,4
上海市农业科技项目(沪农科创字[2022第4-1号])上海市农业科学院卓越团队建设项目(沪农科卓[2022]015)
评论