基于改进YOLOv11的羊只检测方法研究OA
Research on a Sheep Detection Method Based on Improved YOLOv11 Module
[目的]随着深度学习技术的不断发展,目标检测在畜牧业中的应用逐渐受到广泛关注.为提升羊只目标检测的精度与鲁棒性,提出一种基于改进YOLOv11模型的羊只目标检测方法,旨在克服传统检测模型在计算效率和精度方面的不足.[方法]通过引入DynamicConv模块替代传统下采样结构,在保持计算复杂度FLOPs不变的条件下优化参数效率,并通过自适应卷积技术动态调整输入特征,显著提升了多尺度目标检测能力.进一步引入CBAM注意力机制,增强模型对关键特征的感知能力.[结果]实验表明,改进模型在羊只检测任务中,Precision、Recall和mAP50分别较基线模型提升了1.1%、2.8%和0.3%.[结论]实验证明该方法有效提升了羊只目标检测的性能.
[Objective]With the continuous advancement of deep learning technologies,the application of object detection in the field of animal husbandry has garnered increasing attention.For the purpose of enhancing the accuracy and robustness of sheep detection,an improved YOLOv11-based object detection method for sheep was proposed,aiming to overcome the limitations of traditional detection models in terms of computational efficiency and precision.[Methods]The proposed approach integrated a DynamicConv module in place of conventional downsampling structure,optimizing parameter efficiency without increasing the computational complexity(FLOPs).Additionally,adaptive convolution technique was employed to dynamically adjust input features,significantly improving multi-scale object detection capability.The CBAM attention mechanism was further incorporated to enhance the model's sensitivity to key features.[Results]As a result,the improved model achieved increases of 1.1%,2.8%,and 0.3%in Precision,Recall,and mAP50,respectively,compared to the baseline.[Conclusion]The proposed method effectively enhanced the performance of object detection of sheep.
万玉辉;陈新文;左晓佳;赛迪古丽·赛买提;叶尔兰·谢尔毛拉;靳晟
新疆农业大学计算机与信息工程学院,新疆 乌鲁木齐 830052新疆畜牧科学院畜牧业质量标准研究所,新疆 乌鲁木齐 830000新疆畜牧科学院畜牧业质量标准研究所,新疆 乌鲁木齐 830000新疆畜牧科学院畜牧业质量标准研究所,新疆 乌鲁木齐 830000新疆畜牧科学院畜牧业质量标准研究所,新疆 乌鲁木齐 830000新疆农业大学计算机与信息工程学院,新疆 乌鲁木齐 830052
农业科技
羊只检测YOLOv11目标检测动态卷积注意力机制
sheep detectionYOLOv11object detectiondynamic convolutionattention mechanism
《草食家畜》 2026 (2)
64-70,7
新疆维吾尔自治区重点研发计划项目"数字畜牧业关键技术研究与开发"(2023B02013-2)
评论