Högskolan i Skövde

his.sePublications
Planned maintenance
A system upgrade is planned for 10/12-2024, at 12:00-13:00. During this time DiVA will be unavailable.
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Content extraction from marketing flyers
University of Skövde, School of Informatics.
2022 (English)Independent thesis Advanced level (degree of Master (One Year)), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Shops, supermarkets, and real estate dealers use marketing flyers in abundance to advertise the weekly and seasonal offers these days. The information available in such flyers is a good source marketing study. However, this information is not recorded in any central repository for future usage. This work focuses on the feasibility of using deep learning technology to detect objects from marketing flyers. The detection and recognition of objects from media files such as images is a prominent computer vision domain. Although previous investigations have used two-stage object detection techniques to solve the problem using models like Faster R-CNN, this work experiments with the usage of the state-of-the-art single-stage object detecting method YOLO in the object detection from marketing flyers. The work utilizes different training techniques of the YOLO algorithm and identifies the best one to use for detecting objects from marketing flyers.

Transfer learning and custom object training are the two methods of training YOLO. Transfer learning uses the pre-trained knowledge, while custom training is done with two different methodologies. One is by using a pre-annotated dataset like google open images. Another option is to collect representative data and manually annotate the object's position in it. Custom training with manual annotation achieved a mean average precision MAP of 98.56%. Hence it shows that single-stage object detection can be used to detect and classify objects from flyers provided to have representative datasets for training the model.

Place, publisher, year, edition, pages
2022. , p. 2, 22
National Category
Information Systems
Identifiers
URN: urn:nbn:se:his:diva-21536OAI: oai:DiVA.org:his-21536DiVA, id: diva2:1680034
Subject / course
Informationsteknologi
Educational program
Data Science - Master’s Programme
Supervisors
Examiners
Available from: 2022-07-02 Created: 2022-07-02 Last updated: 2022-07-07Bibliographically approved

Open Access in DiVA

fulltext(1015 kB)1117 downloads
File information
File name FULLTEXT02.pdfFile size 1015 kBChecksum SHA-512
27508041f160df544b924a7dcca425050916f26363ae62660c50952f231cdd9490a6831a359e96b0fa4238c1b015fd297ec4ea116d7928db50c5e52253805ebb
Type fulltextMimetype application/pdf

By organisation
School of Informatics
Information Systems

Search outside of DiVA

GoogleGoogle Scholar
Total: 1118 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 471 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf