Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Apr 15;25(8):2477.
doi: 10.3390/s25082477.

Construction and Enhancement of a Rural Road Instance Segmentation Dataset Based on an Improved StyleGAN2-ADA

Zhixin Yao  1   2   3 , Renna Xi  1   2   3 , Taihong Zhang  1   2   3 , Yunjie Zhao  1   2   3 , Yongqiang Tian  1   2   3 , Wenjing Hou  1
Affiliations

Construction and Enhancement of a Rural Road Instance Segmentation Dataset Based on an Improved StyleGAN2-ADA

Zhixin Yao et al. Sensors (Basel). .

Abstract

With the advancement of agricultural automation, the demand for road recognition and understanding in agricultural machinery autonomous driving systems has significantly increased. To address the scarcity of instance segmentation data for rural roads and rural unstructured scenes, particularly the lack of support for high-resolution and fine-grained classification, a 20-class instance segmentation dataset was constructed, comprising 10,062 independently annotated instances. An improved StyleGAN2-ADA data augmentation method was proposed to generate higher-quality image data. This method incorporates a decoupled mapping network (DMN) to reduce the coupling degree of latent codes in W-space and integrates the advantages of convolutional networks and transformers by designing a convolutional coupling transfer block (CCTB). The core cross-shaped window self-attention mechanism in the CCTB enhances the network's ability to capture complex contextual information and spatial layouts. Ablation experiments comparing the improved and original StyleGAN2-ADA networks demonstrate significant improvements, with the inception score (IS) increasing from 42.38 to 77.31 and the Fréchet inception distance (FID) decreasing from 25.09 to 12.42, indicating a notable enhancement in data generation quality and authenticity. In order to verify the effect of data enhancement on the model performance, the algorithms Mask R-CNN, SOLOv2, YOLOv8n, and OneFormer were tested to compare the performance difference between the original dataset and the enhanced dataset, which further confirms the effectiveness of the improved module.

Keywords: StyleGAN; data augmentation; image generation; instance segmentation; rural road.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

Figure 1
Figure 1
Image collection scenario.
Figure 2
Figure 2
Statistical chart of instance category quantities.
Figure 3
Figure 3
Example comparison of original images and mask Images.
Figure 4
Figure 4
Structure diagram of the improved StyleGAN-ALL network.
Figure 5
Figure 5
CCTB module structure diagram.
Figure 6
Figure 6
Cross-shaped window self-attention structure diagram.
Figure 7
Figure 7
Loss and trend diagram of various metrics.
Figure 8
Figure 8
Iterative training effect diagram.
Figure 9
Figure 9
Visualization of OneFormer model.
Figure 10
Figure 10
Visualization of ablation experiment results.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Kabir M., Jim J.R., Istenes Z. Terrain detection and segmentation for autonomous vehicle navigation: A state-of-the-art systematic review. Inf. Fusion. 2025;113:102644. doi: 10.1016/j.inffus.2024.102644. - DOI
    1. Yao Z., Zhao C., Zhang T. Agricultural machinery automatic navigation technology. iScience. 2024;27:108714. doi: 10.1016/j.isci.2023.108714. - DOI - PMC - PubMed
    1. Charisis C., Argyropoulos D. Deep learning-based instance segmentation architectures in agriculture: A review of the scopes and challenges. Smart Agric. Technol. 2024;8:100448. doi: 10.1016/j.atech.2024.100448. - DOI
    1. Lee D.H., Park H.Y., Lee J. A Review on Recent Deep Learning-Based Semantic Segmentation for Urban Greenness Measurement. Sensors. 2024;24:2245. doi: 10.3390/s24072245. - DOI - PMC - PubMed
    1. Liu W., Qiao X., Zhao C., Deng T., Yan F. VP-YOLO: A human visual perception-inspired robust vehicle-pedestrian detection model for complex traffic scenarios. Expert Syst. Appl. 2025;274:126837. doi: 10.1016/j.eswa.2025.126837. - DOI

LinkOut - more resources