Please use this identifier to cite or link to this item:

Можете відсканувати цей QR-код телефоном( програмою "Сканер QR-кодів" ) для збереження.

Title: Application of deep learning in apparel design
Authors: Свірук, Л.
Курочка, С.
Захаркевич, Оксана Василівна
Кулешова, Світлана Геннадіївна
Sviruk, Lesya
Kurochka, Snizhana
Zakharkevich, Oksana
Kuleshova, Svetlana
Keywords: deep learning;garment type;training examples;clothing design
Issue Date: 2017
Citation: Application of deep learning in apparel design / L. Sviruk, S. Kurochka, O. Zakharkevich, S. Kuleshova // International Conference on Technics, Technologies and Education (ICTTE 2017), October 19-20st, 2017. – Yambol, Bulgaria, 2017. – P. 83-91.
Abstract: Deep learning has many applications in contemporary world, which include the tasks of predicting the fashion as well as analysing the images of outfits and forming the textual description of the sketch. The paper is devoted to the examination of fashion search engines that probably could be utilised in apparel industry. Based on the results of performed online survey, it was determined that it is advisable to use as labels for the images specific features of the particular garment type rather than its name. Analysis of images of outerwear, which were obtained with the help of reverse image search, was conducted and images of women's duffle coats were selected. The selected material was sampled for the categorical principal components analysis and general assessment of differences of the garments images. It was determined that less than 30% of original population of images were sampled with required level of accuracy. All found images of the duffle coats cause the similar consumers' impressions. It was determined that deep learning can be used only for the purposes of similarity search for the online shops, which are oriented to the consumers’ impressions rather than to retrieving the features of fashionable outfits.
Content type: Стаття
Appears in Collections:Кафедра технології та конструювання швейних виробів

Files in This Item:
File Description SizeFormat 
sviruk full paper.pdf751,46 kBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.