RGB-Event Fusion for Robust Lane Detection


Jingtao Dong (Beijing Institute of Technology), Hao Zhuang (Beijing Institute of Technology), Hao Yang (Beijing Institute of Technology), Liyuan Pan (Beijing Institute of Technology)
The 35th British Machine Vision Conference

Abstract

Lane detection is crucial for autonomous driving, yet standard RGB cameras struggle under extreme lighting conditions. Event cameras, with their high temporal resolution and low latency, offer a promising solution in such environments, motivating the fusion of RGB and event data to enhance lane detection robustness. To address this challenge, we propose a pipeline incorporating the Multiple Feature Fusion Blocks (MFFB), which leverages the temporal correlation of lane markings by maintaining the current state and providing crucial information for subsequent timesteps. Additionally, we integrate edge features to mitigate the insufficiency of lane edge information caused by insufficient motion in event cameras. Despite the availability of event camera datasets for autonomous driving, no existing dataset simultaneously includes both event and RGB data with annotation for lane detection. To bridge this gap, we introduce DSEC-Lane, the first dataset to provide synchronized RGB frames and event data, capturing autonomous driving scenes under extreme lighting conditions across urban, suburban, and rural environments. Extensive experiments demonstrate that our approach surpasses SOTA methods, highlighting the effectiveness of event-image fusion in challenging lane detection scenarios.

Citation

@inproceedings{Dong_2025_BMVC,
author    = {Jingtao Dong and Hao Zhuang and Hao Yang and Liyuan Pan},
title     = {RGB-Event Fusion for Robust Lane Detection},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_370/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection