What is it about?
We have designed a new real-time semantic segmentation network, FBRNet, which achieves a better balance between segmentation accuracy and speed.
Featured Image
Why is it important?
Our designed arASPP module can extract multiscale features in a short time, with smaller weights and higher segmentation accuracy. The CSFM module further extracts features from both contextual and detailed perspectives, helping to better integrate features from different layers. The LABRM module combines the Laplacian operator and spatial attention mechanism for the first time, fully utilizing the capabilities of existing edge information extraction algorithms, and improving the segmentation accuracy of the model.
Perspectives
Semantic segmentation is one of the classic tasks of computer vision, with a wide range of application scenarios, such as graphics processing, autonomous driving, and medical image analysis, and its task is to segment the image produced in a specific scene.
Shaojun Qu
Hunan Normal University
Read the Original
This page is a summary of: FBRNet: a feature fusion and border refinement network for real-time semantic segmentation, Pattern Analysis and Applications, January 2024, Springer Science + Business Media,
DOI: 10.1007/s10044-023-01207-2.
You can read the full text:
Contributors
The following have contributed to this page
