Efficient Residual Road Condition Segmentation Network Based on Reconstructed Images

Authors: Xiang Shijie, Zhou Dong, Tian Dan

Abstract:

This paper focuses on the application of real-time semantic segmentation technology in complex road condition recognition, aiming to address the critical issue of how to improve segmentation accuracy while ensuring real-time performance. Semantic segmentation technology has broad application prospects in fields such as autonomous vehicle navigation and remote sensing image recognition. However, current real-time semantic segmentation networks face significant technical challenges and optimization gaps in balancing speed and accuracy. To tackle this problem, this paper conducts an in-depth study and proposes a Guided Image Reconstruction Module. By resampling high-resolution images into a set of low-resolution images, this module effectively reduces computational complexity, allowing the network to more efficiently extract features within limited resources, thereby improving the performance of real-time segmentation tasks. In addition, a dual-branch network structure is designed in this paper to fully leverage the advantages of different feature layers. A Hybrid Attention Mechanism is also presented, which can dynamically capture multi-scale contextual information and effectively enhance the focus on important features, thus improving the segmentation accuracy of the network in complex road condition. Compared with traditional methods, the proposed model achieves a better balance between accuracy and real-time performance and demonstrates competitive results in road condition segmentation tasks, showcasing its superiority. Experimental results show that this method not only significantly improves segmentation accuracy while maintaining real-time performance, but also remains stable across diverse and complex road conditions, making it highly applicable in practical scenarios. By incorporating the Guided Image Reconstruction Module, dual-branch structure, and Hybrid Attention Mechanism, this paper presents an approach to real-time semantic segmentation tasks, which is expected to further advance the development of this field.

Keywords: Hybrid attention mechanism, image reconstruction, real-time, road condition recognition.

References:

[1] Paszke A, Chaurasia A, Kim S, et al. Enet: A deep neural network architecture for real-time semantic segmentation (J). 2016.
[2] Romera E, Alvarez J M, Bergasa L M, et al. Erfnet: Efficient residual factorized convnet for real-time semantic segmentation (J). 2017, 19(1): 263-72.
[3] Chen L-C, Papandreou G, Kokkinos I, et al. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs (J). 2017, 40(4): 834-48.
[4] Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network; proceedings of the Proceedings of the IEEE conference on computer vision and pattern recognition, F, 2017 (C).
[5] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation; proceedings of the Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18, F, 2015 (C). Springer.
[6] Tomar N K, Jha D, Riegler M A, et al. Fanet: A feedback attention network for improved biomedical image segmentation (J). 2022, 34(11): 9375-88.
[7] Li G, Yun I, Kim J, et al. Dabnet: Depth-wise asymmetric bottleneck for real-time semantic segmentation (J). 2019.
[8] Yu C, Wang J, Peng C, et al. Bisenet: Bilateral segmentation network for real-time semantic segmentation; proceedings of the Proceedings of the European conference on computer vision (ECCV), F, 2018 (C).
[9] Yu C, Gao C, Wang J, et al. Bisenet v2: Bilateral network with guided aggregation for real-time semantic segmentation (J). 2021, 129: 3051-68.
[10] Lei X, Lu L, Jiang Z, et al. STDC‐MA network for semantic segmentation (J). 2022, 16(14): 3758-67.
[11] Han W, Zhang Z, Zhang Y, et al. Contextnet: Improving convolutional neural networks for automatic speech recognition with global context (J). 2020.
[12] Khanh T L B, Dao D-P, Ho N-H, et al. Enhancing U-Net with spatial-channel attention gate for abnormal tissue segmentation in medical imaging (J). 2020, 10(17): 5729.
[13] Poudel R P, Liwicki S, Cipolla R J a p a. Fast-scnn: Fast semantic segmentation network (J). 2019.
[14] Jenkins M D, Carr T A, Iglesias M I, et al. A Deep Convolutional Neural Network for Semantic Pixel-Wise Segmentation of Road and Pavement Surface Cracks; proceedings of the 2018 26th European Signal Processing Conference (EUSIPCO), F 3-7 Sept. 2018, 2018 (C).
[15] Yadav S, Patra S, Arora C, et al. Deep CNN with color lines model for unmarked road segmentation; proceedings of the 2017 IEEE International Conference on Image Processing (ICIP), F 17-20 Sept. 2017, 2017 (C).
[16] Lan M, Zhang Y, Zhang L, et al. Global context based automatic road segmentation via dilated convolutional neural network (J). Information Sciences, 2020, 535: 156-71.
[17] Wang H, Chen Y, Cai Y, et al. SFNet-N: An improved SFNet algorithm for semantic segmentation of low-light autonomous driving road scenes (J). 2022, 23(11): 21405-17.
[18] Sun X, Xie Y, Jiang L, et al. DMA-Net: DeepLab with Multi-Scale Attention for Pavement Crack Segmentation (J). IEEE Transactions on Intelligent Transportation Systems, 2022, 23(10): 18392-403.
[19] Lin G, Milan A, Shen C, et al. Refinenet: Multi-path refinement networks for high-resolution semantic segmentation; proceedings of the Proceedings of the IEEE conference on computer vision and pattern recognition, F, 2017 (C).
[20] Yurtkulu S C, Şahin Y H, Unal G. Semantic segmentation with extended DeepLabv3 architecture; proceedings of the 2019 27th Signal Processing and Communications Applications Conference (SIU), F, 2019 (C). IEEE.
[21] Wang H, Jiang X, Ren H, et al. Swiftnet: Real-time video object segmentation; proceedings of the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, F, 2021 (C).
[22] Nirkin Y, Wolf L, Hassner T. Hyperseg: Patch-wise hypernetwork for real-time semantic segmentation; proceedings of the Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, F, 2021 (C).
[23] Verelst T, Tuytelaars T J I T o P A, Intelligence M. SegBlocks: Block-based dynamic resolution networks for real-time segmentation (J). 2022, 45(2): 2400-11.
[24] Taghavi P, Langari R, Pandey G J a p a. SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Images (J). 2024.