An Efficienti Fabric Pattern Classification Using Transfer Learning Convolutional Network

EasyChair Preprint 7372

Abstract

Since, fabric classification is a challenging task in the image classification domain, helping business people to quickly and efficiently categorize fabric patterns is very much needed. Many of the online fashion portals uses various machine learning or deep learning model for automated tagging or automated image classification of images in their websites. In recent years transfer learning techniques are very helpful for building task specific models from pretrained networks thus it has received a lot of attraction among researchers and data scientist. This paper proposes an efficient Transfer Learning Convolutional Network (TLCN) for detecting the fabric pattern classes. The proposed system modifies the pretrained network architecture in terms of pooling and convolution operations to make an optimized and a robust model. Four different networks were trained on a dataset which is scrapped from many websites. Experimental analysis showed that VGG16 model performed better than other transfer learning models.

Keyphrases: Convolutional Neural Networks, Fabric classification, Transfer Learning

@booklet{EasyChair:7372,
  author    = {Thomaskutty Reji and Jossy George and Athishay Kesan and Andrea Rodrigues},
  title     = {An Efficienti Fabric Pattern Classification Using Transfer Learning Convolutional Network},
  howpublished = {EasyChair Preprint 7372},
  year      = {EasyChair, 2022}}