基于无人机图像检测的林业虫害监控预防
Forest pest monitoring and prevention based on UAV image detection
邱雅林 1刘向龙 2何小军 3赵庆龙 1贾存芳1
作者信息
- 1. 庆阳市林业和草原科学技术推广站,甘肃 庆阳 745000
- 2. 庆阳市林业科学研究所,甘肃 庆阳 745099
- 3. 子午岭林管局合水分局北川林场,甘肃 庆阳 745400
- 折叠
摘要
为解决现有基于无人机图像的虫害监控方法效率低效果差,且需要耗费大量人力物力的问题,研究基于深度学习构建了基于无人机的林业虫害检测框架,将浅层网络提取的特征信息传递到深层网络,并通过剪枝和批量归一化折叠对模型进行了轻量化改进.结果表明,训练过程中各模型趋于稳定时,改进后的YOLOv4模型平均准确率达97.38%,计算成本和存储需求较原始的YOLOv4已分别降低17.81个百分点和23.38%;平均检测正确率比改进前高12.75个百分点.
Abstract
In order to solve the problem of low efficiency and poor effect of existing pest control methods,which required a lot of man-power and material resources,the research built a forest pest detection framework based on deep learning,which transferred the fea-ture information extracted from the shallow network to the deep network,and made lightweight improvements to the model through pruning and batch normalization folding.The results showed that,when each model tended to be stable during training,the average ac-curacy of the improved YOLOv4 model reached 97.38%,and compared with the original YOLOv4 model,the computing cost and stor-age requirements were reduced by 17.81 percent points and 23.38%,respectively.The average detection accuracy was 12.75 percent points higher than before.
关键词
无人机/虫害监控/图像检测/YOLOv4模型Key words
unmanned aerial vehicle(UAV)/pest control/image detection/YOLOv4 model引用本文复制引用
基金项目
中央财政林业科技推广示范项目(甘[2023]ZYTG 007号)
庆阳市科技计划项目(QY-STK-2022A-042)
出版年
2024
NETL
NSTL
万方数据