<p>This paper presents a comprehensive review of the latest advancements in fabric defect detection leveraging machine learning techniques. It introduces the characteristics of textile defects and public datasets, and encapsulates the challenges encountered in defect detection. A total of 2 000 references spanning diverse time periods are utilized to cluster keywords, revealing that machine learning and deep learning have emerged as the top two. By employing the systematic literature review methodology, it explores the research question, search progress, inclusion and exclusion criteria, and eligibility standards on the basis of the two keywords (machine learning and deep learning). Machine learning algorithms are classified into traditional machine learning and deep learning. The selected references are systematically subdivided into supervised, unsupervised, and semisupervised methods in accordance with the labeled dataset. A thorough analysis of classical network architectures is conducted using public datasets and other datasets, and each network type is accompanied with illustrative examples. The methods are further evaluated on the basis of key metrics in terms of detection success rate, datasets used, detection methodologies, time efficiency, and hardware. This comprehensive evaluation highlights the strengths and limitations of machine learning, particularly deep learning, in textile defect detection. Potential future developments and research directions in this field are explored, offering valuable insights for researchers and engineers working in textile defect detection.</p>