miun.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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
Optimization algorithm selection for object detection and segmentation with Mask R-CNN
Mid Sweden University, Faculty of Science, Technology and Media, Department of Information Systems and Technology.
2019 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Deep learning is a field within machine learning that has grown in popularity. It is used in areas such as: image classification, speech recognition, market price predictions, object detection and much more. The main objective of this study has been to, on the requests of a company, train a model using deep learning to be able to classify and produce masks of objects of interest within images. A comparison of different optimization algorithms was done in order to identify the optimal one for the task at hand. Pixel-wise annotations of the objects were produced in order to train the model. By altering the code of Matterports implementation of Mask R-CNN to train on the dataset (of images) provided by HIAB, the goals were achieved. The optimization algorithm best suited for the conditions of this study was concluded to be AdaGrad. This was concluded based on the mean value of the total loss for each optimization algorithm. In future work, the dataset would preferably be larger in order to increase the predictive quality of the model.

Place, publisher, year, edition, pages
2019. , p. 55
Keywords [en]
deep learning, optimization algorithms, object detection, Mask R-CNN
National Category
Software Engineering
Identifiers
URN: urn:nbn:se:miun:diva-36438Local ID: DT-V19-G3-001OAI: oai:DiVA.org:miun-36438DiVA, id: diva2:1328039
Subject / course
Computer Engineering DT1
Educational program
Master of Science in Engineering - Computer Engineering TDTEA 300 higher education credits
Supervisors
Examiners
Available from: 2019-06-20 Created: 2019-06-20 Last updated: 2019-08-22Bibliographically approved

Open Access in DiVA

No full text in DiVA

Search in DiVA

By author/editor
Yildirim, Kani
By organisation
Department of Information Systems and Technology
Software Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 8 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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