A small fishing vessel recognition method using transfer learning based on laser sensors

文献类型： 外文期刊

第一作者： Zheng, Jianli

作者： Zheng, Jianli;Cao, Jianjun;Yuan, Kun;Liu, Yang

作者机构： Chinese Acad Fishery Sci, Fishery Machinery & Instrument Res Inst, Shanghai 201606, Peoples R China;Liaoning Normal Univ, Sch Comp & Informat Technol, Dalian 116081, Peoples R China

期刊名称：SCIENTIFIC REPORTS （ 2022影响因子：4.6； 五年影响因子：4.9 ）

ISSN： 2045-2322

年卷期： 2023 年 13 卷 1 期

页码：

收录情况： SCI

摘要： The management of small vessels has always been key to maritime administration. This paper presents a novel method for recognizing small fishing vessels based on laser sensors. Using four types of small fishing vessels as targets, a recognition method for small fishing vessels based on Markov transition field (MTF) time-series images and VGG-16 transfer learning is proposed. In contrast to conventional methods, this study uses polynomial fitting to obtain the contours of a fishing vessel and transforms one-dimensional vessel contours into two-dimensional time-series images using the MTF coding method. The VGG-16 model is used for the recognition process, and migration learning is applied to improve the results. The UCR time-series public dataset is used as a transfer learning dataset for the MTF time-series image encoding. The experiment demonstrates that the proposed method exhibits higher accuracy and performance than 1D-CNN and other general neural network models, and the highest accuracy rate is 98.92%.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序