Deep learning-based technique for detecting fast lanes

Zongfeng Liu; Junlong Li; Baoyun Wang

doi:10.1117/12.2673624

22 May 2023 Deep learning-based technique for detecting fast lanes

Zongfeng Liu, Junlong Li, Baoyun Wang

Proceedings Volume 12640, International Conference on Internet of Things and Machine Learning (IoTML 2022); 126401F (2023) https://doi.org/10.1117/12.2673624
Event: International Conference on Internet of Things and Machine Learning (IoTML 2022), 2022, Harbin, China

Abstract

This research suggests a fast lane line identification approach based on deep learning to address the issue of vehicle occlusion in lane line recognition as well as the poor speed of lane line detection. The backbone network's feature extraction speed is increased with the usage of a deep separable convolution technique. In order to address the issue of insufficient lane line recognition when the vehicle is obstructed, the Feature Pyramid Networks (FPN) approach is utilized to improve the extraction of contextual information from the network. To accomplish quick and precise lane line recognition, the lane line thin structure characteristic is completely leveraged, the lane line a priori approach is applied, and the line IoU (L-IoU) idea is used to introduce line IoU loss. The accuracy rate on the Tusimple dataset achieves 0.9652 on the CULane dataset 76.11 F1 score, and the detection speed is 212.5 Fps.

Citation Download Citation

Zongfeng Liu, Junlong Li, and Baoyun Wang "Deep learning-based technique for detecting fast lanes", Proc. SPIE 12640, International Conference on Internet of Things and Machine Learning (IoTML 2022), 126401F (22 May 2023); https://doi.org/10.1117/12.2673624

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available