PulseExploreJournal ClubTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 14, 2026Journal of King Saud University - Computer and Information SciencesOpen Access

Hierarchical vision Mamba with adaptive multi-scale fusion for steel surface defect classification

View Full Paper
Ask AI
Bookmark
Share

Authors

XYXinyi YuHSHaotian SunJPJinmin Peng

Discussion

Loading...

Member takes

Overview

This analysis demonstrates improved defect classification in steel surfaces, indicating better quality control in industrial applications.

Key Points

  • The aim is to enhance steel surface defect classification by improving feature extraction and detection accuracy.
  • Developed a hierarchical architecture, HiAM-Mamba, with adaptive multi-scale processing.
  • Integrated a Multi-Scale Fusion (MSF) module for feature aggregation.
  • Implemented a Gated CNN for capturing fine-grained patterns.
  • Introduced an Adaptive Feature Recalibration mechanism to reduce noise.
  • Utilized a hierarchical head for comprehensive decision-making.
  • Achieved state-of-the-art accuracy of 99.90% and 99.15% on NEU-CLS and X-SDD benchmarks, respectively.
  • Significantly outperformed current CNN and Transformer-based methods.
  • Maintained superior computational efficiency compared to existing approaches.

Cite This Study

Yu et al. (2026) studied this question.

synapsesocial.com/papers/69b4fbb1b39f7826a300c13ehttps://doi.org/10.1007/s44443-026-00623-8
View Full Paper
Ask AI
Bookmark
Share