Related Experiment Video
Updated: May 13, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
434
Lane line detection based on cross-convolutional hybrid attention mechanism
Jianping Wen1, Zhuang Zhao2, Chenze Wang1
1College of Mechanical Engineering, Xi'an University of Science and Technology, Xi'an, 710054, China.
Scientific Reports
|May 9, 2025
Summary
This study introduces CCHA-Net, a novel lane line detection model using a hybrid attention mechanism for improved accuracy in complex driving conditions. The model enhances feature recognition and robustness in dynamic environments.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Machine Learning
Background:
- Lane line recognition is crucial for autonomous driving systems.
- Existing methods struggle with accuracy and robustness in dynamic, complex environments.
Purpose of the Study:
- To propose a novel lane line detection model, CCHA-Net, for enhanced accuracy and robustness.
- To improve lane feature continuity and long-range dependency capture in challenging scenarios.
Main Methods:
- Developed a cross-convolutional hybrid attention mechanism (CCHA-Net) integrating channel and spatial attention.
- Implemented cross-group feature interaction and dynamic spatial weight allocation.
- Designed a lightweight message-passing module with dual-branch multi-scale convolutions for feature fusion and parameter reduction.
Main Results:
- CCHA-Net achieved an F1 score of 80.2% on the CULane dataset.
- The model attained an accuracy of 96.8% on the TuSimple dataset.
- Demonstrated significant enhancement in lane line recognition accuracy in complex environments.
Conclusions:
- The proposed CCHA-Net model effectively improves lane line detection accuracy and robustness.
- The integrated attention mechanism and lightweight design offer a promising solution for real-world autonomous driving applications.
