Institutional Repository of Vilnius University
Beždžionių raupų sukeltų odos pakitimų atpažinimas naudojantis giliojo mokymosi modeliais /
Abstract
dc:descriptionMonkeypox causes a variety of skin lesions that are unique. The most common way to record the spread of this virus is when a sick person transmits the virus to another person. Deep neural networks have been used for a long time in the field of skin disease recognition. In order to have an initial diagnosis of monkeypox, deep neural networks have been used to visually detect skin lesions caused by this disease. The first attempts to identify skin lesions caused by monkeypox were made using the VGG16 deep learning model. The VGG16 deep learning model showed promising results in recognising monkeypox virus from photographs, achieving an accuracy of 83%. When comparing the previous studies with the MSLD dataset, the best results were obtained with the MobileNetV2 deep learning model, which achieved 91.11% accuracy. Since the topic is quite new and there is little research in this area, this work decided to improve recognition by testing the EfficientNetV2 series models together with the traditional ResNet50, VGG16, InceptionV3 models and applying average and majority voting ensemble methods to the best performing models. This study was able to improve the recognition of skin lesions caused by monkeypox by fitting the EfficientNetV2 models with the majority vote ensemble method, achieving an accuracy of 93.34% and maintaining high metrics of precision, recall, and F1–score.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Rinkevičius, Aivaras,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192847262&prefLang=en_US