MFANet: Mixed Feature Attention Network for Port Lane Detection

Jinwei Zhang, Jinya Su, Dewei Yi, Jun Yang

Research output: Chapter in Book/Report/Conference proceedingPublished conference contribution

Abstract

With the advancement of industrial automation and artificial intelligence technology, unmanned port autonomy has gained increasing attention. Port automatic driving technology is a critical component of unmanned autonomous systems, enabling rubber tire gantry cranes (RTGs) to autonomously navigate by detecting port lanes and implementing control commands. In this study, a deep learning-based method is proposed for detecting port lanes in complex port scenes. The method employs the Inception module and attention mechanism to enhance lane feature extraction, and a structural loss is introduced to explicitly constrain the detection results and incorporate the prior characteristics of port lanes. A novel Mixed Feature Attention Network (MFANet) is proposed to implement the method for port lane detection. Experimental results demonstrate that MFANet effectively improves the accuracy of port lane detection while maintaining high computational efficiency. Furthermore, the performance of MFANet is evaluated in a real-world application scenario.

Original languageEnglish
Title of host publication2023 42nd Chinese Control Conference, CCC 2023
PublisherIEEE Computer Society
Pages8633-8638
Number of pages6
ISBN (Electronic)9789887581543
DOIs
Publication statusPublished - 18 Sept 2023
Event42nd Chinese Control Conference, CCC 2023 - Tianjin, China
Duration: 24 Jul 202326 Jul 2023

Publication series

NameChinese Control Conference, CCC
Volume2023-July
ISSN (Print)1934-1768
ISSN (Electronic)2161-2927

Conference

Conference42nd Chinese Control Conference, CCC 2023
Country/TerritoryChina
CityTianjin
Period24/07/2326/07/23

Keywords

  • Inception
  • Mixed Attention mechanism
  • Port lane detection
  • Shape loss

Fingerprint

Dive into the research topics of 'MFANet: Mixed Feature Attention Network for Port Lane Detection'. Together they form a unique fingerprint.

Cite this