What question did this study set out to answer?

The aim is to enhance concrete crack segmentation accuracy using a lightweight CNN–Mamba hybrid network for varied inspection conditions.

March 27, 2026ElectronicsOpen Access

Lightweight CNN–Mamba Hybrid Network for Multi-Scale Concrete Crack Segmentation Using Vision Sensors

View Full Paper

Ask AI

Bookmark

Authors

Demonstrates improved crack segmentation in concrete using a hybrid CNN–Mamba network, suggesting effective maintenance support.

The aim is to enhance concrete crack segmentation accuracy using a lightweight CNN–Mamba hybrid network for varied inspection conditions.
Developed a lightweight CNN–Mamba hybrid segmentation framework based on Vm-unet.
Utilized boundary-sensitive convolutional features and long-range state-space representations.
Implemented a decoder interaction fusion scheme to improve crack continuity and boundary definition.
Tested on a multi-source composite dataset and public benchmarks.
Achieved 80.11% mean Intersection over Union (mIoU) and 82.05% Dice score on the composite dataset.
Showcased consistent improvements over CNN, Transformer, and Mamba-based methods.
Maintained efficient performance with 36.049 GFLOPs and 25.991 million parameters.

Guan et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

March 27, 2026ElectronicsOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Demonstrates improved crack segmentation in concrete using a hybrid CNN–Mamba network, suggesting effective maintenance support.

The aim is to enhance concrete crack segmentation accuracy using a lightweight CNN–Mamba hybrid network for varied inspection conditions.
Developed a lightweight CNN–Mamba hybrid segmentation framework based on Vm-unet.
Utilized boundary-sensitive convolutional features and long-range state-space representations.
Implemented a decoder interaction fusion scheme to improve crack continuity and boundary definition.
Tested on a multi-source composite dataset and public benchmarks.
Achieved 80.11% mean Intersection over Union (mIoU) and 82.05% Dice score on the composite dataset.
Showcased consistent improvements over CNN, Transformer, and Mamba-based methods.
Maintained efficient performance with 36.049 GFLOPs and 25.991 million parameters.

Guan et al. (2026) studied this question.

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark