Detection of Fabric Defects Using Convolutional Networks

Defects on fabric surface, which greatly affects product quality, is an important issue under investigation for engineers, researchers, and organizations working in respective sectoral industry or academia. Increasing demand and growing supply capacity cause problems such as time and labor during and after the production phase. In addition, the limited nature of the human causes time loss and low quality of the fabric. The loss rate in fabric quality observations based on human control has been shown to be 40%. In order to reduce this rate, automated systems should be activated in fabric quality inspection. In recent years, many studies have been carried for this purpose. With the developing technology, correct fabric defect detection is about 100% success rate. Especially though the use of new trend approaches (mostly learning based algorithms), the correct detection of defect has been moved to a very advanced level. In the present paper, VGG19 is applied to classify defects correctly. VGG19 produced high performance outputs.