Automated Fabric Defect Detection

EasyChair Preprint 12723

Abstract

This paper discussed an automated system for detecting fabric defects, which is a state-ofthe-art solution to the problems associated with manual fabric inspection in the textile industry. The need for automated, dependable, and efficient quality control systems is increasing in tandem with the ongoing transformation of production processes. Traditional manual inspection methods are laborious and subjective, which leads to uneven defect detection. Defect identification is inconsistent due to the subjective nature and lengthy processing times of traditional hand inspection methods. Using sophisticated algorithms, the system initially examines high-resolution pictures of fabric samples in order to optimize characteristics and minimize variations in lighting and fabric textures. Using the ResNet architectures, the two CNN models created in this work had average accuracies of 89.84% and 93.45%, respectively, indicating statistically significant findings.

Keyphrases: Classification, Convolutional Neural Network, Identification of Fabric Defect, Textile Industry, image processing, machine learning

@booklet{EasyChair:12723,
  author    = {Pushpanjali Sajjanshetti and Chaitanya Patil and Marshal Naik and Ganesh Waghmare and Sudarshan Gaikwad},
  title     = {Automated Fabric Defect Detection},
  howpublished = {EasyChair Preprint 12723},
  year      = {EasyChair, 2024}}