LAKASABrowsing by Author "Bejar, Shane"

Now showing 1 - 1 of 1

Validating inception V3 for signiture verification
(Don Mariano Marcos Memorial State University - South La Union Campus, 2023-05) Pacho, Domingo R.; Bejar, Shane; Dulay, Jenly M.; Ignacio, Danica J.
This paper presents is aimed at validating the effectiveness of the Inception V3 model in signature verification by applying preprocessing and augmentation technique. Experimental research was utilized where in experimentation, collection, and analysis of numerical and non-numerical data were performed The datasets of Signature acquired from the CEDAR datasets. Specifically, it sought answers to attain the following objectives: (1) apply preprocessing and augmentation technique to CEDAR dataset, (2) develop a signature verification model using Inception V3, and (3) evaluate the performance of the signature verification model. The dataset underwent preprocessing and augmentation techniques. The signature verification model s performance was measured using the model straining s loss and accuracy compared to the validation s loss and accuracy respectively. Furthermore, FRR, FAR is also employed to measure the performance of the model. The SGD optimizer with a batch size of 32 yields the best results among the other parameters having an accuracy of 97. 10%.