Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 12, 2026International Journal of Computational Intelligence SystemsOpen Access

CMD-CrackNet: A Modular, Interpretable, and Lightweight Approach to Crack Segmentation under Data Scarcity

View Full Paper
Ask AI
Bookmark
Share

Authors

YGYu GanSSS Muhammad Ahmed Hassan ShahAAAbdullah I Al-Mansour

Discussion

Loading...

Member takes

Overview

Randomized trial evaluates crack segmentation in infrastructure monitoring, suggesting better efficiency under data scarcity.

Key Points

  • The aim is to enhance pavement crack segmentation using a lightweight framework under limited labeled data conditions.
  • Developed CMD-CrackNet incorporating AttnCLR for feature generalization from unlabeled data.
  • Implemented TriDecoderNet with three distinct decoders for tailored predictions and context modeling.
  • Used a Selective Channel-Spatial Enhancement (SCSE) module to boost efficiency while maintaining a compact design.
  • Achieved 96.8% pixel accuracy with 1.04 million parameters.
  • Obtained a 74.74% Dice score, outperforming existing deeper baseline models.
  • Generated Monte Carlo uncertainty maps and Grad-CAM visualizations for improved interpretability.

Cite This Study

Gan et al. (2026) studied this question.

synapsesocial.com/papers/6a2ba5068101cf8926f033e9https://doi.org/10.1007/s44196-026-01430-9
View Full Paper
Ask AI
Bookmark
Share

Also Consider

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

  1. 1CrackNet: A novel multi-scale architecture for crack segmentation2026 · 1 citations
  2. 2Enabling Real-Time, Cost-Efficient, and Lightweight High-Speed Crack Segmentation using Self-Supervised Attention Mechanism2026
  3. 3Scalable and Accurate Crack Segmentation for Infrastructure Health Monitoring Using EfficientNetB0 and a Context-Aware Attention Mechanism2026 · 1 citations
  4. 4CSegNet：A Crack Segmentation Network Combining CNN and Transformer2024 · 5 citations
  5. 5Efficient CrackUNet: Hierarchical Spatial-Channel Attention with Multi-Scale Fusion for Pavement and Bridge Crack Segmentation2025 · 1 citations