摘要
海草的检测与识别在海洋资源研究、海洋科普、海产品加工、非物质文化遗产保护等领域有着重要的意义。为确保对渤海地区的海草资源进行有效的分类与识别,本文基于Keras框架,以inception v3网络模型为基础,采用迁移学习,结合特征融合优化分类,从而对渤海地区的十种珍贵海草进行有效识别,结果显示正确率达到了92.53%。
The detection and identification of seaweed is of great significance in the fields of marine resources research,marine science popularization,marine products processing,intangible cultural heritage protection and so on.In order to ensure the effective classification and recognition of seagrass resources in the Bohai Sea area,based on keras framework and the inception V3 network model,this paper uses feature fusion to optimize the classification to effectively identify the ten precious seagrass resources in the Bohai Sea area.The results showthat the accuracy reaches 93.53%.
作者
孙婷婷
陈林林
SUN tingting;CHEN linlin(Weihai Ocean Vocational College,Shandong Rongcheng 264300,China;Weihai intelligent marine fishery equipment engineering technology research center,Shandong Rongcheng 264300,China)
出处
《智能计算机与应用》
2020年第7期314-317,共4页
Intelligent Computer and Applications
