International Journal of Image, Graphics and Signal Processing (IJIGSP)
IJIGSP Vol. 18, No. 1, 8 Feb. 2026
Cover page and Table of Contents: PDF (size: 1731KB)
Multi-Scale and Auxiliary-Supervised Architectures for Accurate Road Network Mapping
PDF (1731KB), PP.51-69
Views: 0 Downloads: 0
Author(s)
Index Terms
Road Extraction, Satellite Imagery, Deep Learning, Multi-Scale Architecture, Loss Function
Abstract
Automated road network extraction from satellite imagery represents a critical advancement for Geographic Information Systems (GIS) applications in infrastructure management and urban planning. This paper introduces two novel deep learning architectures based on LinkNet: RoadNet-MS (Multi-Scale) and RoadNet-AUX (Multi-Scale with Auxiliary Supervision), specifically designed to enhance road segmentation performance. RoadNet-MS incorporates Multi-Scale Contextual Blocks (CMS-Blocks) and hybrid blocks to effectively capture diverse contextual features at multiple scales, achieving F1-scores of 78.87% on the challenging DeepGlobe dataset and 82.30% on the Boston & Los Angeles dataset. RoadNet-AUX extends this architecture through auxiliary supervision, further improving performance with F1-scores of 79.14% on DeepGlobe and 82.33% on Boston-LA. Both proposed architectures demonstrate competitive performance and consistent improvements over existing methods, including the state-of-the-art NL-LinkNet, across both evaluation datasets. Notably, RoadNet-MS achieves the highest precision (83.55%) among all compared methods on DeepGlobe. These contributions provide a pathway toward more accurate and scalable road network mapping, essential for modern urban planning and infrastructure monitoring applications.
Cite This Paper
Mohamed El Mehdi Imam, Lila Meddeber, Tarik Zouagui, "Multi-Scale and Auxiliary-Supervised Architectures for Accurate Road Network Mapping", International Journal of Image, Graphics and Signal Processing(IJIGSP), Vol.18, No.1, pp. 51-69, 2026. DOI:10.5815/ijigsp.2026.01.04
Reference
[1]N. N. Kourgialas et al., “A web-based GIS platform supporting innovative irrigation management techniques at farm-scale for the Mediterranean island of Crete,” Sci. Total Environ., vol. 842, p. 156918, Oct. 2022, doi: 10.1016/j.scitotenv.2022.156918.
[2]L. Meddeber, T. Zouagui, and N. Berrached, “Efficient photometric and geometric stitching approach for remote sensing images based on wavelet transform and local invariant,” J. Appl. Remote Sens., vol. 15, no. 03, Jul. 2021, doi: 10.1117/1.JRS.15.034502.
[3]S. Mo, Y. Shi, Q. Yuan, and M. Li, “A Survey of Deep Learning Road Extraction Algorithms Using High-Resolution Remote Sensing Images,” Sensors, vol. 24, no. 5, p. 1708, Mar. 2024, doi: 10.3390/s24051708.
[4]H. R. R. Bakhtiari, A. Abdollahi, and H. Rezaeian, “Semi automatic road extraction from digital images,” Egypt. J. Remote Sens. Space Sci., vol. 20, no. 1, pp. 117–123, Jun. 2017, doi: 10.1016/j.ejrs.2017.03.001.
[5]O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” May 18, 2015, arXiv: arXiv:1505.04597. doi: 10.48550/arXiv.1505.04597.
[6]A. Chaurasia and E. Culurciello, “LinkNet: Exploiting encoder representations for efficient semantic segmentation,” in 2017 IEEE Visual Communications and Image Processing (VCIP), St. Petersburg, FL: IEEE, Dec. 2017, pp. 1–4. doi: 10.1109/VCIP.2017.8305148.
[7]Y. Wang, J. Seo, and T. Jeon, “NL-LinkNet: Toward Lighter But More Accurate Road Extraction With Nonlocal Operations,” IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, 2022, doi: 10.1109/LGRS.2021.3050477.
[8]W. Wang, N. Yang, Y. Zhang, F. Wang, T. Cao, and P. Eklund, “A review of road extraction from remote sensing images,” J. Traffic Transp. Eng. Engl. Ed., vol. 3, no. 3, pp. 271–282, Jun. 2016, doi: 10.1016/j.jtte.2016.05.005.
[9]S. Gupta, S. Suman, S. Kumar Sharma, P. K. Garg, and S. K. Ghosh, “Development of web GIS based accident information system for safe and sustainable transport,” Mater. Today Proc., p. S2214785323043523, Aug. 2023, doi: 10.1016/j.matpr.2023.08.098.
[10]M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: Active contour models,” Int. J. Comput. Vis., vol. 1, no. 4, pp. 321–331, Jan. 1988, doi: 10.1007/BF00133570.
[11]N. S. Kumar, B. Sukanya, B. Mohan, and G. Prathibha, “Extraction of Roads from Satellite Images Based on Edge Detection,” vol. 5, no. 2, 2017.
[12]A. Di Benedetto, M. Fiani, and L. M. Gujski, “U-Net-Based CNN Architecture for Road Crack Segmentation,” Infrastructures, vol. 8, no. 5, p. 90, May 2023, doi: 10.3390/infrastructures8050090.
[13]N. Y. Q. Abderrahim, S. Abderrahim, and A. Rida, “Road Segmentation using U-Net architecture,” in 2020 IEEE International conference of Moroccan Geomatics (Morgeo), Casablanca, Morocco: IEEE, May 2020, pp. 1–4. doi: 10.1109/Morgeo49228.2020.9121887.
[14]S. Xie, R. Girshick, P. Dollár, Z. Tu, and K. He, “Aggregated Residual Transformations for Deep Neural Networks,” Apr. 10, 2017, arXiv: arXiv:1611.05431. Accessed: May 26, 2024. [Online]. Available: http://arxiv.org/abs/1611.05431
[15]Z. Fan, Y. Liu, M. Xia, J. Hou, F. Yan, and Q. Zang, “ResAt-UNet: A U-Shaped Network Using ResNet and Attention Module for Image Segmentation of Urban Buildings,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 16, pp. 2094–2111, 2023, doi: 10.1109/JSTARS.2023.3238720.
[16]L. Zhou, C. Zhang, and M. Wu, “D-LinkNet: LinkNet with Pretrained Encoder and Dilated Convolution for High Resolution Satellite Imagery Road Extraction,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, UT, USA: IEEE, Jun. 2018, pp. 192–1924. doi: 10.1109/CVPRW.2018.00034.
[17]S. Li and X. Liu, “Multi-type road extraction and analysis of high-resolution images with D-LinkNet50,” in 2022 3rd International Conference on Geology, Mapping and Remote Sensing (ICGMRS), Zhoushan, China: IEEE, Apr. 2022, pp. 244–248. doi: 10.1109/ICGMRS55602.2022.9849390.
[18]X. Wang, R. Girshick, A. Gupta, and K. He, “Non-local Neural Networks,” Apr. 13, 2018, arXiv: arXiv:1711.07971. Accessed: May 27, 2024. [Online]. Available: http://arxiv.org/abs/1711.07971
[19]Z. Li, H. Chen, N. Jing, and J. Li, “RemainNet: Explore Road Extraction from Remote Sensing Image Using Mask Image Modeling,” Remote Sens., vol. 15, no. 17, p. 4215, Aug. 2023, doi: 10.3390/rs15174215.
[20]Y. Zhang, L. Zhang, Y. Wang, and W. Xu, “AGF-Net: adaptive global feature fusion network for road extraction from remote-sensing images,” Complex Intell. Syst., vol. 10, no. 3, pp. 4311–4328, Jun. 2024, doi: 10.1007/s40747-024-01364-9.
[21]X. Li, S. Yang, F. Meng, W. Li, Z. Yang, and R. Wei, “LCMorph: Exploiting Frequency Cues and Morphological Perception for Low-Contrast Road Extraction in Remote Sensing Images,” Remote Sens., vol. 17, no. 2, p. 257, Jan. 2025, doi: 10.3390/rs17020257.
[22]X. Wang, C. Qin, M. Bai, Q. Ma, and G. Li, “CAFormer: a connectivity-aware vision transformer for road extraction from remote sensing images,” Vis. Comput., vol. 41, no. 10, pp. 7965–7981, Aug. 2025, doi: 10.1007/s00371-025-03849-1.
[23]F. Li, S. Guo, F. Han, J. Zhao, and F. Shen, “Multi-Scale Dilated Convolution Network for Long-Term Time Series Forecasting,” May 14, 2024, arXiv: arXiv:2405.05499. doi: 10.48550/arXiv.2405.05499.
[24]B. Xu, X. Liu, and L. Zhang, “Dilated convolution neural operator for multiscale partial differential equations,” Jul. 16, 2024, arXiv: arXiv:2408.00775. doi: 10.48550/arXiv.2408.00775.
[25]Z. Zhao, P. Ma, M. Jia, X. Wang, and X. Hei, “A Dilated Convolutional Neural Network for Cross-Layers of Contextual Information for Congested Crowd Counting,” Sensors, vol. 24, no. 6, p. 1816, Mar. 2024, doi: 10.3390/s24061816.
[26]J. Su et al., “Momentum Auxiliary Network for Supervised Local Learning,” Aug. 12, 2024, arXiv: arXiv:2407.05623. doi: 10.48550/arXiv.2407.05623.
[27]F. Yu and V. Koltun, “Multi-Scale Context Aggregation by Dilated Convolutions,” Apr. 30, 2016, arXiv: arXiv:1511.07122. Accessed: Nov. 19, 2024. [Online]. Available: http://arxiv.org/abs/1511.07122
[28]A. Wulamu, Z. Shi, D. Zhang, and Z. He, “Multiscale Road Extraction in Remote Sensing Images,” Comput. Intell. Neurosci., vol. 2019, pp. 1–9, Jul. 2019, doi: 10.1155/2019/2373798.
[29]X. Li, L. Lei, and G. Kuang, “Multilevel Adaptive-Scale Context Aggregating Network for Semantic Segmentation in High-Resolution Remote Sensing Images,” IEEE Geosci. Remote Sens. Lett., vol. 19, pp. 1–5, 2022, doi: 10.1109/LGRS.2021.3091284.
[30]R. Zhang et al., “D-FusionNet: road extraction from remote sensing images using dilated convolutional block,” GIScience Remote Sens., vol. 60, no. 1, p. 2270806, Dec. 2023, doi: 10.1080/15481603.2023.2270806.
[31]S. Zagoruyko and N. Komodakis, “Wide Residual Networks,” Jun. 14, 2017, arXiv: arXiv:1605.07146. Accessed: Jun. 03, 2024. [Online]. Available: http://arxiv.org/abs/1605.07146
[32]A. Galdran, G. Carneiro, and M. Á. G. Ballester, “On the Optimal Combination of Cross-Entropy and Soft Dice Losses for Lesion Segmentation with Out-of-Distribution Robustness,” Sep. 14, 2022, arXiv: arXiv:2209.06078. Accessed: May 20, 2024. [Online]. Available: http://arxiv.org/abs/2209.06078
[33]I. Demir et al., “DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Images,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, UT, USA: IEEE, Jun. 2018, pp. 172–17209. doi: 10.1109/CVPRW.2018.00031.
[34]T. Laborie, “Road Segmentation - Boston & Los Angeles.” Accessed: Jan. 06, 2025. [Online]. Available: https://www.kaggle.com/datasets/timothlaborie/roadsegmentation-boston-losangeles
[35]D. Powers, “Evaluation: From Precision, Recall and F-Factor to ROC, Informedness, Markedness & Correlation”.
[36]H. Rezatofighi, N. Tsoi, J. Gwak, A. Sadeghian, I. Reid, and S. Savarese, “Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression,” Apr. 14, 2019, arXiv: arXiv:1902.09630. Accessed: Jun. 27, 2024. [Online]. Available: http://arxiv.org/abs/1902.09630