Tools
Tools
Rajendran, Krishnachander (2020) Performance optimization on skin lesion image classification using pre-trained models. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (816kB) | Preview |
Preview |
PDF (Configuration manual)
Download (3MB) | Preview |
Abstract
Skin disease classification through CNN has become more sophisticated with the inception of high resolution training image datasets. The automatic classification of skin diseases act as the much needed alternative for the traditional methods such as biopsy and cutaneous examination. The requirement of feature extraction is also not necessary with the CNN as opposed to the automated skin classifiers that were built in the past. This shift in focus is toward CNN enabled the introduction of pre-trained models to be used on image classification as part of transfer learning. Such an approach is followed in this research with the use of MobileNet model and DenseNet201 models along with a custom built CNN. All three models are pitted against each other for performance evaluation with the main criteria being the optimizer being used. Adam optimizer and Stochastic Gradient Descent (SGD) are used to explore the hypothesis that SGD optimizer performs better than Adam optimizer. The accuracy obtained from the research provides evidence that the MobileNet model with the SGD optimizer performs better (91.68 %) than the other models in terms of categorical accuracy.
| Item Type: | Thesis (Masters) |
|---|---|
| Subjects: | Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Computer software T Technology > T Technology (General) > Information Technology > Computer software R Medicine > R Medicine (General) |
| Divisions: | School of Computing > Master of Science in Data Analytics |
| Depositing User: | Dan English |
| Date Deposited: | 10 Jun 2020 15:11 |
| Last Modified: | 10 Jun 2020 15:11 |
| URI: | https://norma.ncirl.ie/id/eprint/4263 |
Actions (login required)
 |
View Item |