Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 27, 2025Open Access

Enhanced Crack Segmentation via Dual-Branch CNN-Transformer Architecture with Linear Perception and Multi-Scale Refinement

View Full Paper
Ask AI
Bookmark
Share

Authors

JFJiangtao FengBeijing Institute of TechnologyJLJing LiYanshan UniversityFSFeng SuNanjing University

Discussion

Loading...

Member takes

Overview

Dueling CNN and Transformer models improve crack segmentation performance in infrastructure monitoring, suggesting greater accuracy.

Key Points

  • The model achieves superior crack segmentation, evident from an F1-score of 82.0% on the CrackVision12K dataset.
  • Model utilizes a novel dual-branch architecture that integrates CNN and transformer for enhanced feature extraction.
  • Includes advancements like linear perception guidance and multi-scale edge refinement to boost crack detection accuracy.
  • Demonstrates significant improvements over existing methods, particularly in low-contrast and fine-grained environments.

Cite This Study

Feng et al. (2025) studied this question.

synapsesocial.com/papers/68d7be5eeebfec0fc523777dhttps://doi.org/10.21203/rs.3.rs-7703561/v1
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. 1Directional multi-scale CNN-transformer hybrid network for robust crack segmentation2026
  2. 2Hybrid Multi-Scale CNN-Transformer Network for Structural Surface Crack Segmentation2026 · 2 citations
  3. 3A Dual-Path CNN and Transformer Network for Continuous Pavement Crack Detection2026
  4. 4CrackNet: A novel multi-scale architecture for crack segmentation2026 · 1 citations
  5. 5PCTC-Net: A Crack Segmentation Network with Parallel Dual Encoder Network Fusing Pre-Conv-Based Transformers and Convolutional Neural Networks2024 · 5 citations