EUVIMEDEUVIMED entdecken
Crossref

EMFFTrans: Efficient Multi-Scale Feature Fusion Transformer for Road Scene Semantic Segmentation

Yaocong Hu, Pindeng Wang, Mengbo Jia, Yifan Zhang, Jinwen Hong, Guoyang Wan, Ke Xu, Huicheng Yang, Bingyou Liu, Chao Xie, Tianming Ni, Teng Li

IEEE Transactions on Intelligent Transportation Systems · 2026-06 · 27 (6) · 6589-6604

DOI: 10.1109/tits.2026.3679203 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Intelligent Transportation Systems, 2026-06. Verfasst von Yaocong Hu und 11 weiteren Autor:innen. Die Arbeit nennt 89 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Anhui Higher Education Institutions of China, Open Research Fund of Anhui Provincial Key Laboratory of Advanced Detection and Intelligent Perception.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Road scene semantic segmentation is one of the most crucial tasks in autonomous driving. Unlike general semantic segmentation, traffic applications require fast image processing, making it difficult to satisfy both high-performance and real-time. Moreover, how to efficiently fuse multi-scale feature representations for a comprehensive understanding of the road scene remains challenging. To address these issues, we propose a lightweight U-shape deep learning framework for high-performance and real-time road scene semantic segmentation in this paper. Specifically, we design the lightweight mixed convolutional block (LMCB), which combines depthwise separable convolution and asymmetric dilated convolution, with the purpose of reducing computational cost and expanding the receptive field for road scene feature extraction. Then, efficient multi-scale feature fusion Transformer (EMFFTrans) is proposed to bridge the encoder-decoder interaction, leveraging the separable cross-channel attention (SCCA) mechanism to capture the inherent dependencies across multi-scale feature maps, thereby facilitating comprehensive representation for road scenes. Experimental results show that our proposed framework, running on a single RTX 3090 GPU card, strikes a satisfactory balance between accuracy and efficiency. Specifically, the LMCB-EMFFTrans-Slim model achieves 73.7% mIOU at 166 FPS on the CamVid dataset and 76.3% mIOU at 113 FPS on the Cityscapes dataset with fewer than 1 million learnable parameters. The code will be soon available at: https://github.com/WH-IPUSLab/EMFFTrans

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Anhui Higher Education Institutions of China
  • Open Research Fund of Anhui Provincial Key Laboratory of Advanced Detection and Intelligent Perception
  • Open Foundation of State Key Laboratory of High-End Compressor and System Technology
  • 2024 Anhui Province Intelligent Robot Information Fusion and Control Engineering Research Center Open Project
  • Key Research and Development and Achievement Transformation Projects of Wuhu Science and Technology Plan

Zitieren

APA

Hu, Y., Wang, P., Jia, M., Zhang, Y., Hong, J., Wan, G., Xu, K., Yang, H., Liu, B., Xie, C., Ni, T., & Li, T. (2026). EMFFTrans: Efficient Multi-Scale Feature Fusion Transformer for Road Scene Semantic Segmentation. IEEE Transactions on Intelligent Transportation Systems, 27(6), 6589-6604. https://doi.org/10.1109/tits.2026.3679203

Vancouver

Hu Y, Wang P, Jia M, Zhang Y, Hong J, Wan G, et al. EMFFTrans: Efficient Multi-Scale Feature Fusion Transformer for Road Scene Semantic Segmentation. IEEE Transactions on Intelligent Transportation Systems. 2026;27(6):6589-6604. doi:10.1109/tits.2026.3679203

BibTeX
@article{Hu2026,
  author = {Hu, Yaocong and Wang, Pindeng and Jia, Mengbo and Zhang, Yifan and Hong, Jinwen and Wan, Guoyang and Xu, Ke and Yang, Huicheng and Liu, Bingyou and Xie, Chao and Ni, Tianming and Li, Teng},
  title = {EMFFTrans: Efficient Multi-Scale Feature Fusion Transformer for Road Scene Semantic Segmentation},
  journal = {IEEE Transactions on Intelligent Transportation Systems},
  year = {2026},
  volume = {27},
  number = {6},
  pages = {6589--6604},
  doi = {10.1109/tits.2026.3679203}
}
Kostenlos starten →