Application Of Artificial Neural Networks In Fabric Design Identification.

Summary

In this work, an approach to apply one of the artificial intelligence techniques in the
textile industry is presented. An evaluation of the accuracy of using artificial neural
networks in classifying different weave designs, based on the principle of the back
propagation of the learning rule, is given. The neural network is trained to classify the
different fabric designs representing those most commonly used in the weaving
industry. Through applying this technique a step forward to achieving a complete
quick response system can be accomplished. The weave design can be introduced as
a hand draft or a fabric image by using a scanner. The weave design is then
introduced to the neural network in a matrix form for training and classification. This
method can be used to create an easy archive of the different designs and to activate
the various mechanisms of the loom to operate an electronic dobby or jacquard used
on the modern looms. The results obtained using the above-mentioned technique
prove that artificial neural networks are capable of accurate identification of various
fabric designs.

This paper was first published on the earlier Egytex website. The text above is the summary that was published there.