Fabric Defect Detection using Deep Learning

The detection of fabric defects is critical in terms of industry quality. Many ways are being tried to achieve high precision using image processing studies in order to develop automatic systems for fault detection. Deep learning, which separates from multi-layer architectures and demonstrates great achievement, is applied to fabric defect identification in this study. Auto encoder, a deep learning method aiming at representing input data through compression or decompression, is used to detect fabric defects and achieves satisfactory success. The primary purpose of this study is to improve feature extraction performance by fine-tuning the auto encoder's input value and hyper parameters.

Keywords: deep learning, fabric defect detection, auto encoder, feature extraction