A hybrid attention multi-scale fusion network for real-time semantic segmentation
2025

A hybrid attention multi-scale fusion network for real-time semantic segmentation

publication Evidence: high

Author Information

Author(s): Ye Baofeng, Wu Qianlong

Primary Institution: Qiqihar University, China

Hypothesis

The study proposes a new method to improve semantic segmentation accuracy by addressing the loss of spatial information in existing algorithms.

Conclusion

The proposed method achieves a mean Intersection over Union (mIoU) of 73.6% at 176 frames per second on the Cityscapes dataset, demonstrating improved accuracy and speed.

Supporting Evidence

  • The proposed method achieves 73.6% mIoU at 176 FPS on the Cityscapes dataset.
  • Edge detection methods were incorporated to enhance boundary information extraction.
  • The study addresses the loss of spatial information in existing semantic segmentation algorithms.

Takeaway

This study created a new way to help computers understand images better by keeping important details while still being fast.

Methodology

The study designed two new modules, HFRM and HFFM, to enhance feature extraction and fusion in semantic segmentation tasks, incorporating edge detection methods.

Digital Object Identifier (DOI)

10.1038/s41598-024-84685-6

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication