基于FCN网格定位和特征融合的水培生菜幼苗状态检测方法

Detection Method of Hydroponic Lettuce Seedlings Status Based on FCN Grid Location and Feature Fusion

为了及时发现问题幼苗状态和提高幼苗分拣效率,以水培生菜幼苗培育过程中出现的死亡和双株状态为研究对象,提出一种基于FCN网格定位和特征融合的水培生菜幼苗状态检测方法。在原有研究的基础上,针对双株状态幼苗检测精度低的问题,引入FCN架构改变原有边框回归方式,利用其对位置信息敏感的特性,获取精确的网格点空间信息。同时,利用特征融合策略,充分获取不同网格点间的相关性,实现对水培生菜幼苗问题状态的精准定位。试验结果表明,该方法的平均检测精度达到88.1%,检测精度优于原有方法、FSAF、YOLOv3、FoveaBox、ATSS和CornerNet,尤其对双株状态的幼苗检测精度得到明显提升。该方法能够实现水培生菜问题幼苗状态的自动检测,为水培蔬菜育苗分拣智能化及种植自动化提供技术支持。

In order to find problematic seedlings status timely and improve sorting efficiency of seedlings in cultivation stage of hydroponic vegetables,an automatic detection method of hydroponic lettuce seedlings status based on FCN grid location and feature fusion was proposed,taking dead and double-planting status of seedlings growing in a hole as research object.Aiming at problem of low detection accuracy of two-plant seedlings status,FCN architecture was introduced to change traditional localization based on regression and adopted its sensitive characteristics to obtain accurate grid point spatial information on basis of previous research.At the same time,feature fusion strategy was used to fully obtain correlation between different grid points,so as to achieve further accurate location of problematic status of hydroponic lettuce seedlings.Experimental results showed that mean average precision of this method was 88.1%,which was higher than that of previous method,FSAF,YOLOv3,FoveaBox,ATSS and CornerNet.In particular,detection accuracy of twoplant seedlings status was significantly increased.Hydroponic lettuce seedling condition detection method proposed could realize automatic detection of problem status of hydroponic lettuce seedlings,and provide technical support for intelligent breeding and sorting of hydroponic vegetable seedlings and planting automation.Therefore,the method proposed could realize accurate identification and localization automatically,which could provide technical support for intelligent sorting and automatic planting of hydroponic vegetable seedlings.