安徽农业科学2025，Vol.53Issue(7)：211-218,8.DOI:10.3969/j.issn.0517-6611.2025.07.039
基于改进YOLOv8s模型的河蟹幼苗雌雄检测方法
Gender Detection Method for Crabs Seedlings Based on Enhanced YOLOv8s Model
李显 1马明 1胡志刚 1王可 1计闯 1蒋亚军 1付丹丹1
作者信息
- 1. 武汉轻工大学机械工程学院,湖北武汉 10496
- 折叠
摘要
Abstract
In response to the problem of unclear differentiation between males and females during the seedling stage of river crabs,and insuffi-cient accuracy in manual classification of male and female in river crab seedlings.This article proposes a method for detecting the gender of river crab seedlings based on an improved YOLOv8 model.The improvement method is as follows:firstly,replace the fourth layer C2f module in the backbone network with C2f GAM(global attention,mechanism,GAM)module,adjust the weight of feature information,and reduce the loss of feature information during transmission;secondly,replace the original boundary loss function with the WIoUv1(weighted interpolation of sequential evidence for intersection over union v1)boundary loss function to enhance the anchor box quality during the prediction process and improve the model's generalization ability;finally,the nearest neighbor interpolation upsampling method in the head network is replaced with CARAFE(content aware reassembly of features)upsampling method,which gives the model a larger receptive field and improves its perform-ance.The experimental results show that the accuracy,recall,and average precision of the improved model are 98.4%,91.1%,and 96.1%,respectively,which are 3.2,3.2,and 2.9 percentage point higher than the original model.The results indicate the feasibility of applying ma-chine vision to the classification of male and female crab seedlings and the effectiveness of the improved method.关键词
河蟹幼苗/YOLOv8s/WIoU/GAM/CRAFEKey words
Crab seedlings/YOLOv8s/WIoU/GAM/CRAFE分类
农业科技引用本文复制引用
李显,马明,胡志刚,王可,计闯,蒋亚军,付丹丹..基于改进YOLOv8s模型的河蟹幼苗雌雄检测方法[J].安徽农业科学,2025,53(7):211-218,8.
