Multi-scale object detection algorithm for ship intelligent navigation

被引：0
|
作者：
Xu H. [1 ,2 ]
Long Z. [2 ]
Feng H. [1 ,2 ]
机构：
[1] Key Laboratory of High Performance Ship Technology of Ministry of Education, Wuhan University of Technology, Wuhan
[2] School of Transportation, Wuhan University of Technology, Wuhan
来源：
Huazhong Keji Daxue Xuebao (Ziran Kexue Ban)/Journal of Huazhong University of Science and Technology (Natural Science Edition) | 2021年 / 49卷 / 05期
关键词：
Convolutional neural network; Feature fusion; Intelligent navigation; Object detection; Sample distribution;
D O I：
10.13245/j.hust.210509
中图分类号：
学科分类号：
摘要：
To solve the problems that scene on ship's intelligent navigation is complex and the obstacle scale in the acquired image information varies greatly, a multi-scale object detection algorithm was proposed based on YOLO (you only look once) algorithm. First, an adaptive feature fusion module was designed so that the feature map used for detection had strong semantic information of various scales. Then, a new loss function was designed to alleviate the problem of uneven sample distribution and optimize the training process. Finally, the simulation experiment of water surface image dataset shows that the proposed algorithm has obvious advantages in multi-scale detection, improving the detection accuracy, especially the accuracy of small targets increased by about 34%, without significantly increasing the reasoning time and the number of parameters, and has high real-time performance. © 2021, Editorial Board of Journal of Huazhong University of Science and Technology. All right reserved.
引用
收藏
页码：50 / 55
页数：5
相关论文
共 14 条
  • [1] 1
  • [2] 3
  • [3] KRIZHEVSKY A, SUTSKEVER I, HINTON G E., Imagenet classification with deep convolutional neural networks, Advances in Neural Information Processing Systems, 25, 2, pp. 1097-1105, (2012)
  • [4] 1
  • [5] 49, 4, pp. 38-43
  • [6] REDMON J, DIVVALA S, GIRSHICK R, Et al., You only look once: unified, real-time object detection, Proc of Computer Vision and Pattern Recognition, pp. 779-788, (2015)
  • [7] (2018)
  • [8] REDMON J, FARHADI A., YOLO V3: an incremental improvement, Proc of IEEE Conference on Computer Vision and Pattern Recognition, pp. 89-95, (2018)
  • [9] (2019)
  • [10] LIU W, ANGUEIOU D, ERHAN D, Et al., SSD: single shot multibox detector, Proc of European Conference on Computer Vision, pp. 21-37, (2016)
12