Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Classification of IKEA home furnishing items into design style groups using deep learning methods

Gabnyte, Emilija LU (2023) DABN01 20231
Department of Economics
Department of Statistics
Abstract
Accurately categorizing home furnishing items into interior design style groups plays a crucial role in delivering personalized recommendations that align with customers' design style preferences. However, manually labeling items by experts is a costly and resource-intensive task. To address this challenge, deep learning methods are used to automate the manual labeling of IKEA home furnishing items. The first part of the analysis considers descriptions of the items and classifies them into interior design style groups existing within the company. Leveraging the information present in the descriptions shows the potential for predicting the style group of previously unlabeled and new items. The second part of the study introduces more... (More)
Accurately categorizing home furnishing items into interior design style groups plays a crucial role in delivering personalized recommendations that align with customers' design style preferences. However, manually labeling items by experts is a costly and resource-intensive task. To address this challenge, deep learning methods are used to automate the manual labeling of IKEA home furnishing items. The first part of the analysis considers descriptions of the items and classifies them into interior design style groups existing within the company. Leveraging the information present in the descriptions shows the potential for predicting the style group of previously unlabeled and new items. The second part of the study introduces more customer-friendly style groups, shifts the focus to the images of the items, and explores a multi-label classification approach. The experimental results reveal challenges posed by the limited size of the dataset and highlight the need for additional data collection and refinement of the models to improve their performance. (Less)
Please use this url to cite or link to this publication:
author
Gabnyte, Emilija LU
supervisor
organization
course
DABN01 20231
year
type
H1 - Master's Degree (One Year)
subject
keywords
Home furnishing, interior design style, deep learning, DistilBERT, ViT, EfficientNet
language
English
id
9118691
date added to LUP
2023-11-21 12:53:49
date last changed
2023-11-21 13:27:26
@misc{9118691,
  abstract     = {{Accurately categorizing home furnishing items into interior design style groups plays a crucial role in delivering personalized recommendations that align with customers' design style preferences. However, manually labeling items by experts is a costly and resource-intensive task. To address this challenge, deep learning methods are used to automate the manual labeling of IKEA home furnishing items. The first part of the analysis considers descriptions of the items and classifies them into interior design style groups existing within the company. Leveraging the information present in the descriptions shows the potential for predicting the style group of previously unlabeled and new items. The second part of the study introduces more customer-friendly style groups, shifts the focus to the images of the items, and explores a multi-label classification approach. The experimental results reveal challenges posed by the limited size of the dataset and highlight the need for additional data collection and refinement of the models to improve their performance.}},
  author       = {{Gabnyte, Emilija}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Classification of IKEA home furnishing items into design style groups using deep learning methods}},
  year         = {{2023}},
}