The Lane Detection Model Based on Data Augmentation and Deep Learning

Yunye Zhu

doi:10.54254/2755-2721/2025.LD26122

Applied and Computational EngineeringOpen access

The Lane Detection Model Based on Data Augmentation and Deep Learning

Research Article

Open Access

The Lane Detection Model Based on Data Augmentation and Deep Learning

Yunye Zhu ^1*

¹ Xi’an Jiao Tong University, Xi’an, China

^*Corresponding author: 1827430139@qq.com

Published on 20 August 2025

ACE Vol.179

ISSN (Print): 2755-273X

ISSN (Online): 2755-2721

ISBN (Print): 978-1-80590-184-6

ISBN (Online): 978-1-80590-129-7

Download Cover

Abstract

Lane detection serves as a cornerstone task in autonomous driving systems, as it directly impacts the vehicle’s ability to maintain lane discipline, ensure safety, and perform accurate path planning. Although U-Net-based deep learning models have demonstrated strong potential for automatic lane segmentation, their performance can degrade significantly under complex real-world conditions such as variable lighting, occlusions, and worn or curved lane markings.To address these limitations, this study proposes an enhanced lane detection framework built upon the U-Net architecture. The proposed model integrates three key improvements: (1) advanced data augmentation techniques to increase the diversity and robustness of the training data, (2) a refined loss function combining PolyLoss and contrastive loss to address foreground-background imbalance and enhance structural learning, and (3) an optimized upsampling strategy designed to better preserve spatial details and lane continuity in the output predictions.Extensive experiments conducted on the TuSimple lane detection benchmark validate the effectiveness of our approach. The enhanced model achieves an Intersection over Union (IoU) of 44.49%, significantly surpassing the baseline U-Net’s performance of 40.36%. These results confirm that the proposed modifications not only improve segmentation accuracy but also enhance the model’s robustness and generalization capability in real-world driving scenarios. Overall, this work contributes practical insights and techniques that can facilitate the deployment of lane detection systems in intelligent transportation and autonomous vehicle platforms.

Keywords:

Lane detection, U-Net, Data augmentation, Loss function optimization, Autonomous driving

View PDF

References

[1]. X. Yang, H. Zhang, and Y. Liu, “A survey on deep learning-based lane detection, ” IEEE Transactions on Intelligent Transportation Systems, 2023.

[2]. J. Zheng, W. Liu, and R. Zhao, “Ccha-net: Cross convolution hybrid attention network for lane detection, ” Scientific Reports, 2024. [Online]. Available: https: //www.nature.com/articles/s41598- 025-01167-z

[3]. M. Li, F. Zhou, and J. Wang, “Learning lane detection via self-attention and hybrid loss, ” Pattern Recognition Letters, vol. 156, pp. 45–52, 2022.

[4]. K. Zhou and S. Huang, “Attentionlane: Lightweight transformer with multi-level guidance for lane detection, ” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023, pp. 11 789–11 797.

[5]. O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation, ” in Medical Image Computing and Computer-Assisted Intervention (MICCAI). Springer, 2015, pp. 234–241.

[6]. R. Yousri, K. Moussa, M. A. Elattar, and A. H. Madian, “A novel data augmentation approach for egolane detection enhancement using perspective transformation, ” Evolving Systems, vol. 15, p. 1021–1032, 2023.

[7]. R. Li and Y. Dong, “Robust lane detection through self pre-training with masked sequential autoencoders and finetuning with customized polyloss, ” arXiv preprint arXiv: 2305.17271, 2023.

[8]. K. Zhou, Y. Feng, and J. Li, “Unsupervised domain adaptive lane detection via contextual contrast and aggregation, ” arXiv preprint arXiv: 2407.13328, 2024.

[9]. “Grid anchor lane detection based on attribute correlation, ” Applied Sciences, 2025.

[10]. K. Xu, Z. Hao, M. Zhu, and J. Wang, “An efficient lane detection network with channelenhanced coordinate attention, ” Machines, vol. 12, no. 12, p. 870, 2024.

[11]. A. Csato and F. Mariasiu, “Effects of hybridizing the unet neural network in traffic lane detection process, ” Applied Sciences, vol. 15, no. 14, p. 7970, 2025.

[12]. M. Tangestanizadeh, M. Dehghani Tezerjani, and S. Youseffan Jazi, “Attentionbased unet method for autonomous lane detection, ” arXiv preprint arXiv: 2411.10902, 2024.

[13]. K. Zhou, Y. Feng, and J. Li, “Unsupervised domain adaptive lane detection via contextual contrast and aggregation, ” arXiv preprint arXiv: 2407.13328, 2024.

[14]. S. S. M. Salehi, D. Erdogmus, and A. Gholipour, “Tversky loss function for image segmentation using 3d fully convolutional deep networks, ” MICCAI, 2017.

[15]. F. Milletari, N. Navab, and S.-A. Ahmadi, “V-net: Fully convolutional neural networks for volumetric medical image segmentation, ” 3DV, 2016.

[16]. M. Berman, A. Triki, and M. B. Blaschko, “The lovász-softmax loss: A tractable surrogate for the optimization of the intersection-over-union measure in neural networks, ” CVPR, 2018.

[17]. W. Li et al., “Boundary-aware network for lane detection, ” IEEE TITS, 2022.

[18]. A. Tarvainen and H. Valpola, “Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results, ” NeurIPS, 2017.

[19]. T. Inc., “Tusimple lane detection dataset, ” 2020, accessed: 2025-07-26. [Online]. Available: https: //www.kaggle.com/datasets/manideep1108/tusimple