A Lightweight Dual-Stream Enhanced Network With Multilevel Feature Guidance and Hybrid Knowledge Distillation for Steel Surface Defect Detection

The advancement of deep neural networks has significantly improved the precision of strip surface defect detection (SDD). However, current methods typically face difficulties in meeting real-time performance and computational efficiency requirements, particularly in high-speed industrial manufacturing environments with limited computational resources. To address these challenges, a dual-stream enhancement strategy is proposed, resulting in a lightweight dual-stream enhanced network (LDENet) that effectively balances detection accuracy, speed, and resource efficiency. Specifically, a lightweight PConv-stacked backbone network is introduced to minimize model complexity and improve inference speed. The Multi-layer Coordinated Adaptive Feature Fusion (MCAF) module reduces feature redundancy by adaptively integrating important information across multiple layers. Moreover, a hybrid knowledge distillation (HKD) method improves the lightweight model's capability by transferring detailed information from a pre-trained teacher model. On the NEU-DET dataset, LDENet achieves 81.1 mAP and 181 fps with only 1.41M parameters, demonstrating its suitability for real-time, resource-constrained industrial applications.