What question did this study set out to answer?

The research aims to develop a lightweight and efficient model for detecting steel surface defects under complex scenarios.

March 25, 2026SensorsOpen Access

LHO-net: A Lightweight Steel Defect Detection Framework Based on Cross-Scale Feature Selection and Adaptive Optimization

View Full Paper

Ask AI

Bookmark

Authors

A novel framework achieves efficient defect detection in steel surfaces, enhancing recognition in resource-limited devices.

The research aims to develop a lightweight and efficient model for detecting steel surface defects under complex scenarios.
Introduced the LHO-net with LM Backbone for dynamic feature fusion.
Utilized HSPAN for multi-scale feature fusion with adaptive upsampling.
Employed OAHead for occlusion compensation via feature aggregation and normalization.
Evaluated performance on NEU-DET and GC-10 datasets.
Achieved mAP@0.5 of 75.0% and recall of 73.6% on NEU-DET dataset.
Reduced parameters by 64% and computational cost by 60.3% compared to YOLOv12.
Demonstrated superior stability in detecting complex defects on the GC-10 dataset.
Maintained stable recognition performance while minimizing false detections.

Wang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark