Global context guided refinement and aggregation network for lightweight surface defect detection

Feng Yan

doi:10.1016/j.patcog.2025.112893

Global context guided refinement and aggregation network for lightweight surface defect detection

2025-12-10

0

PRE

AI

DOI：10.1016/j.patcog.2025.112893

Original

Original request for help

Save

Abstract

En 中文

• A fast and accurate pixel-level surface defect detection network with global guidance is proposed. • A Depthwise Self-Attention is proposed to learn global information with low computation cost. • A Channel Cross-Attention is proposed to help suppress the background interference from low-level features. • The advantages of the proposed approach over existing methods are analyzed.

Global context guided refinement and aggregation network for lightweight surface defect detection

Abstract

Journal

Organization