Skin disease detection using improved bag of features algorithm

College

College of Computer Studies

Department/Unit

Computer Science

Document Type

Conference Proceeding

Source Title

5th Iranian Conference on Signal Processing and Intelligent Systems (ICSPIS)

First Page

1

Last Page

5

Publication Date

12-18-2019

Abstract

This study uses digital image processing to develop a model to detect common skin diseases in the Philippines; acne and BOIL. The researchers used different methods and technique such as; improved bag of features algorithm, speeded up robust features algorithm, interest point detection, Gaussian filtering and k-means clustering. The overall accuracy rate of the system is 96% while overall loss is (0.03), and the total average confidence rate of the tests done with different test data in terms of detection and classification is 98.48%. Moreover, the average precision/recall rate of combined images for the two categories is 99% In the confusion matrix, Acne got the highest number of correct predicted skin disease. On the other hand, BOIL got the lowest number of correctly classified. Besides, Acne got the highest precision result of 98%, while BOIL got a high precision result of 97%. In recall results, both models have the same percentage of 99%.

html

Disciplines

Computer Sciences

Keywords

Image processing—Digital techniques; Skin—Diseases

Upload File

wf_no

This document is currently not available here.

Share

COinS