您好，欢迎访问中国水产科学研究院 机构知识库！
中国水产科学研究院机构知识库

全部

  • 全部
  • 标题
  • 学者
  • 机构
  • 关键词

Deep Learning-Based Fish Detection Using Above-Water Infrared Camera for Deep-Sea Aquaculture: A Comparison Study

文献类型： 外文期刊

作者： Li, Gen 1 ; Yao, Zidan 5 ; Hu, Yu 1 ; Lian, Anji 1 ; Yuan, Taiping 1 ; Pang, Guoliang 1 ; Huang, Xiaohua 1 ;

作者机构： 1.Chinese Acad Fishery Sci, South China Sea Fisheries Res Inst, Guangzhou 510300, Peoples R China

2.Minist Agr & Rural Affairs, Key Lab Open Sea Fishery Dev, Guangzhou 510300, Peoples R China

3.Chinese Acad Fishery Sci, South China Sea Fisheries Res Inst, Res & Dev Ctr Trop Aquat Prod, Sanya 572018, Peoples R China

4.Sanya Trop Fisheries Res Inst, Sanya 572018, Peoples R China

5.Zhejiang Ocean Univ, Sch Marine Engn Equipment, Zhoushan 316022, Peoples R China

关键词： fish detection; fish dataset; Faster R-CNN; above-water infrared camera; deep-sea aquaculture

期刊名称：SENSORS （ 影响因子：3.9； 五年影响因子：4.1 ）

ISSN：

年卷期： 2024 年 24 卷 8 期

页码：

收录情况： SCI

摘要： Long-term, automated fish detection provides invaluable data for deep-sea aquaculture, which is crucial for safe and efficient seawater aquafarming. In this paper, we used an infrared camera installed on a deep-sea truss-structure net cage to collect fish images, which were subsequently labeled to establish a fish dataset. Comparison experiments with our dataset based on Faster R-CNN as the basic objection detection framework were conducted to explore how different backbone networks and network improvement modules influenced fish detection performances. Furthermore, we also experimented with the effects of different learning rates, feature extraction layers, and data augmentation strategies. Our results showed that Faster R-CNN with the EfficientNetB0 backbone and FPN module was the most competitive fish detection network for our dataset, since it took a significantly shorter detection time while maintaining a high AP50 value of 0.85, compared to the best AP50 value of 0.86 being achieved by the combination of VGG16 with all improvement modules plus data augmentation. Overall, this work has verified the effectiveness of deep learning-based object detection methods and provided insights into subsequent network improvements.

  • 相关文献
作者其他论文 更多>>
置顶
购物车
反馈

© 京ICP备09089781号-29 主办单位：中国水产科学研究院

学术资源： 中国农业科学院 农业专业知识服务系统 农科联盟资源共建共享平台 中国农业科技文献与信息服务平台 中国农业期刊集成服务平台