融合注意力机制的轻量级猪体质量实时估计方法  

Lightweight real-time pig weight estimation method with attention mechanism

作　　者：罗森 葛志辉[1] LUO Sen;GE Zhihui(School of Computer,Electronics and Information,Guangxi University,Nanning 530004,China)

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004

出　　处：《广西大学学报(自然科学版)》2025年第1期161-172,共12页Journal of Guangxi University(Natural Science Edition)

基　　金：国家自然科学基金项目(62371144)。

摘　　要：为了在资源有限的移动设备上对猪体质量进行准确和实时估计,提出了融合注意力机制的轻量级猪体质量实时估计方法,该方法分为分割阶段和体质量估计阶段。在分割阶段,为了能更准确地获得猪躯干分割图像,提出了边缘引导的轻量级分割网络EG-YOLO。在体质量估计阶段,为了更好地融合双视角特征用于体质量估计,基于改进的轻量级MobileNeXt子网络,构建了双视角特征融合的轻量级体质量估计网络,并在自建的猪体质量估计数据集上进行实验。结果表明,体质量估计的误差仅为3.99%。In order to achieve accurate and real-time pig weight estimation on resource-constrained mobile devices,a lightweight pig weight real-time estimation methocls was proposed,which consisted of a segmentation stage and a weight estimation stage.In the segmentation stage,an edge-guided lightweight segmentation network,named EG-YOLO,was presented to obtain more accurate segmentation images of the pig body.In the weight estimation stage,to better fuse dual-view features for weight estimation,a lightweight weight estimation network with dual-view feature fusion was constructed based on an improved lightweight MobileNeXt sub-network.Experimental results on a self-built pig weight estimation dataset show that the weight estimation error is only 3.99%.

关 键 词：体质量估计 边缘引导 注意力机制 特征融合 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]