<p>The textile industry has been one of Taiwan’s key traditional sectors. In the past, it has been labor-intensive and low-tech, but recent advances have led to widespread automation across most production stages. Yarn packages, which serve as intermediate products for yarn transportation, are now produced through automated processes—except for defect inspection, which still relies heavily on manual visual checks. To reduce labor costs and improve inspection quality, this study explores automated defect detection methods aimed at assisting inspectors and enhancing efficiency. As high-performance fabrics require stringent quality control, defect detection has become a critical step in the manufacturing process. We propose two categories of detection methods based on spatial and frequency domain analysis to identify linear defects on yarn-package surfaces. Experimental results show that spatial-domain methods struggle to detect defects whose orientation closely resembles the surrounding yarn texture. In contrast, the two frequency-domain approaches effectively compensate for these limitations, offering improved detection of subtle and texture-aligned defects.</p>