Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
August 14, 2026Discover Applied SciencesOpen Access

Library spine text detection based on SAM and CFA

View Full Paper
Ask AI
Bookmark
Share

Authors

HFHaiping Fan

Discussion

Loading...

Member takes

Overview

Experimental model demonstrates accurate spine text detection via segmentation and attention mechanisms, indicating improved efficiency for smart library automation.

Key Points

  • To develop a robust spine text detection framework by combining a general segmentation model with a cross-feature attention mechanism to improve small-target character detection and feature fusion.
  • Integrated a multi-feature fusion branch into an enhanced general segmentation framework to balance global semantics and local detail extraction.
  • Implemented a cross-feature attention mechanism for dynamic weight allocation and cross-scale feature interaction.
  • Evaluated performance on public datasets and tested environmental robustness under extreme low-light and high-light conditions against baseline detectors (DBNet, PAN, and EAST).
  • Achieved a maximum text detection accuracy of 95.12% on public datasets, alongside an 8–15% increase in mask consistency during ablation testing.
  • Maintained intersection-over-union consistency above 87% and optical character recognition accuracy above 85% in extreme low-light and high-light scenes.
  • Outperformed DBNet, PAN, and EAST by 6.0–8.5 percentage points in localization accuracy and orientation consistency while reducing computational load by approximately 18%.

Cite This Study

Haiping Fan (2026) studied this question.

synapsesocial.com/papers/6a7ec6c6b70b84ec8b912ea2https://doi.org/10.1007/s42452-026-09252-2
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. 1A Multi-Scale Natural Scene Text Detection Method Based on Attention Feature Extraction and Cascade Feature Fusion2024 · 10 citations
  2. 2ASM-DBNet: Introducing Adaptive Differentiable Binarization, Spatial-Channel Self-Attention and Multi-Scale Context-Enhanced Dynamic Upsampling for Natural Scene Text Detection2026
  3. 3A Real‐Time Automated Library Inventory System Based on Edge‐Cloud Collaboration2026
  4. 4Enhancing medical text detection with vision-language pre-training and efficient segmentation2024 · 9 citations
  5. 5A Two-Stage End-to-End Framework for Robust Scene Text Spotting with Self-Calibrated Detection and Contextual Recognition2025 · 4 citations