期刊论文

中文

渔业现代化

1007-9580

2024

51

5

90-101

10.3969/j.issn.1007-9580.2024.05.011

湖北省重点研发计划(2023BBB046)。

本校为第一机构

为提高鲨鱼种群细粒度分类的准确率,解决数据集图像干扰因素多、图像局部关键特征提取不足、通道语义关联欠缺等问题,提出了一种基于改进双线性卷积神经网络B-CNN模型的DM-BCNN鲨鱼种群细粒度分类模型。首先,引入可变形卷积...展开更多 为提高鲨鱼种群细粒度分类的准确率,解决数据集图像干扰因素多、图像局部关键特征提取不足、通道语义关联欠缺等问题,提出了一种基于改进双线性卷积神经网络B-CNN模型的DM-BCNN鲨鱼种群细粒度分类模型。首先,引入可变形卷积将原模型中的特征提取部分替换为DRAM_ResNet网络结构,提升模型对复杂非规则形状和局部结构的检测能力;然后在此基础上采用NAM注意力机制,加...

To enhance the accuracy of fine-grained classification of shark populations and address issues such as image interference,insufficient extraction of local key features,and lack of semantic correlation between channels,a DM-BCNN model for fine-grained classification of s...MORE To enhance the accuracy of fine-grained classification of shark populations and address issues such as image interference,insufficient extraction of local key features,and lack of semantic correlation between channels,a DM-BCNN model for fine-grained classification of sha...