lu.se

Datavetenskap

Lunds Tekniska Högskola

Kalendarium

CS MSc Thesis Zoom Presentation 7 May 2021

Föreläsning

Tid: 2021-05-07 15:15 till 16:00
Plats: Online via: https://lu-se.zoom.us/j/63247367354
Kontakt: birger [dot] swahn [at] cs [dot] lth [dot] se
Spara händelsen till din kalender


One Computer Science MSc thesis to be presented on 7 May via Zoom

Friday, 7 May there will be a master thesis presentation in Computer Science at Lund University, Faculty of Engineering.

The presentation will take place via Zoom at: https://lu-se.zoom.us/j/63247367354

Note to potential opponents: N.B. No more opponents for this presentation (Register as an opponent to the presentation of your choice by sending an email to the examiner for that presentation (firstname.lastname@cs.lth.se). Do not forget to specify the presentation you register for! Note that the number of opponents may be limited (often to two), so you might be forced to choose another presentation if you register too late. Registrations are individual, just as the oppositions are! More instructions are found on this page.)


15:15-16:00

Presenters: Jesper Berg, Emil Wihlander
Title: Finding company logos through their websites: A multimodal approach
Examiner: Flavius Gruian
Supervisors: Marcus Klang (LTH), Frank Camara (AFRY Group AB)

Retrieving information about a company from their website is possible through web-scraping. A good example of this is the logo which has several use-cases. The obstacle however is finding which of all the images is the requested logo. This thesis presents different models for classifying if a web-scaped image is the specified company's logo. These models vary from trivial, machine learning text and image classifiers to multimodal solutions. Each model is trained on a diverse data set with companies and websites. They are thereafter evaluated and compared using several classification scores. Text classification models showed viability in selecting company logos, while image methods had problems with context and filtering out other logos from other companies or brands. Multimodal methods showed high potiential and using a weight-based decision-model performed better than any other method, with an MCC of 0.7042. This indicates multimodality is an encouraging approach to finding company logos.

Link to presentation: https://lu-se.zoom.us/j/63247367354

Link to popular science summary: https://fileadmin.cs.lth.se/cs/Education/Examensarbete/Popsci/210507_15BergWihlander.pdf