arrow
Volume 11, Issue 4
Classification of Female Apparel using Convolutional Neural Network

Qiao-Qi Li, Yue-Qi Zhong & Xin Wang

Journal of Fiber Bioengineering & Informatics, 11 (2018), pp. 209-216.

Published online: 2019-02

Export citation
  • Abstract

With the vigorous development of clothing e-commerce, the amount of clothing image data on the internet has increased dramatically. A tedious effort was required to manually label and classify the semantic attributes of clothing images. Manual marking is time-consuming and laborious, so a method of automatic classification using convolutional neural networks was studied. In this paper, a female cloth dataset consisting of 10 types of female clothing was built. Convolutional Neural Network (CNN) was employed to learn the feature vectors for each type. Five different types of architectures, including ResNet50, Inception-v3, and VGG-19, AlexNet, and FashionNet were used for performance comparison. Experimental results have shown that Inception-v3 possesses the highest accuracy (98.07% for training and 96.91% for testing) in clothing classification compared with other methods.

  • AMS Subject Headings

  • Copyright

COPYRIGHT: © Global Science Press

  • Email address

zhyq@dhu.edu.cn (Yue-Qi Zhong)

  • BibTex
  • RIS
  • TXT
@Article{JFBI-11-209, author = {Li , Qiao-QiZhong , Yue-Qi and Wang , Xin}, title = {Classification of Female Apparel using Convolutional Neural Network}, journal = {Journal of Fiber Bioengineering and Informatics}, year = {2019}, volume = {11}, number = {4}, pages = {209--216}, abstract = {

With the vigorous development of clothing e-commerce, the amount of clothing image data on the internet has increased dramatically. A tedious effort was required to manually label and classify the semantic attributes of clothing images. Manual marking is time-consuming and laborious, so a method of automatic classification using convolutional neural networks was studied. In this paper, a female cloth dataset consisting of 10 types of female clothing was built. Convolutional Neural Network (CNN) was employed to learn the feature vectors for each type. Five different types of architectures, including ResNet50, Inception-v3, and VGG-19, AlexNet, and FashionNet were used for performance comparison. Experimental results have shown that Inception-v3 possesses the highest accuracy (98.07% for training and 96.91% for testing) in clothing classification compared with other methods.

}, issn = {2617-8699}, doi = {https://doi.org/10.3993/jfbim00319}, url = {http://global-sci.org/intro/article_detail/jfbi/13010.html} }
TY - JOUR T1 - Classification of Female Apparel using Convolutional Neural Network AU - Li , Qiao-Qi AU - Zhong , Yue-Qi AU - Wang , Xin JO - Journal of Fiber Bioengineering and Informatics VL - 4 SP - 209 EP - 216 PY - 2019 DA - 2019/02 SN - 11 DO - http://doi.org/10.3993/jfbim00319 UR - https://global-sci.org/intro/article_detail/jfbi/13010.html KW - Female Clothing Image KW - Image Classification KW - Convolutional Neural Network KW - Deep Learning. AB -

With the vigorous development of clothing e-commerce, the amount of clothing image data on the internet has increased dramatically. A tedious effort was required to manually label and classify the semantic attributes of clothing images. Manual marking is time-consuming and laborious, so a method of automatic classification using convolutional neural networks was studied. In this paper, a female cloth dataset consisting of 10 types of female clothing was built. Convolutional Neural Network (CNN) was employed to learn the feature vectors for each type. Five different types of architectures, including ResNet50, Inception-v3, and VGG-19, AlexNet, and FashionNet were used for performance comparison. Experimental results have shown that Inception-v3 possesses the highest accuracy (98.07% for training and 96.91% for testing) in clothing classification compared with other methods.

Qiao-Qi Li, Yue-Qi Zhong & Xin Wang. (2020). Classification of Female Apparel using Convolutional Neural Network. Journal of Fiber Bioengineering and Informatics. 11 (4). 209-216. doi:10.3993/jfbim00319
Copy to clipboard
The citation has been copied to your clipboard