| Abstract: | Ethiopia is known in different Cultural clothing in Fabrics, design and colour based on
Ethnics, Geographical Location and their religions. Peoples of Ethiopia different cultural
clothing is their conventional clothing on occasion of different ceremonial events. To identify
such different cultural cloth it needs human expert this method consumes time and man
labour and for clothes nearly the same fabrics it is difficult to identify with human eye vision.
Previously there was no developed model to identify Ethiopian cultural cloth to overcome
this problem we use deep learning CNN Model to classify selected cultural clothing of
Ethiopia. : Different Ethiopian Cultural clothing image collected from Different area, such as
from Oromo cultural centre, Ethiopian Minister of Culture and tourism, Ethiopian Regional
states media and their cultural centres. We classified image into sixteen way Softmax
classifier was used for categorizing into specific classes (i.e., of Afar, Amhara, Beshangul
Gumuz, Dawro, Gambella, Gurage, Hadiya, Harari, Kaffa, Kambata, Oromo, Sidama, Siltie,
Somali, Tigray and Welayta). We collected 11,200 each class 700 images and from total
image 80% of image used for training and other 20% for validating the Model. Those
collected image resized into equal image size of 224x224 image pixels. After compared CNN
with Three, five and Seven Convulation we get Model CNN Contains Five Convolution
Layer Feature extraction each layer with kernel size 3x3, Maxpooling 2x2 and Bach
normalization, Last Layer with Flatten Layer and to identify the result we used Softmax
activation. Final we get image accuracy of 97.21. This model Recognizes and classify
commonly dressed Ethiopian cultural cloth also This model in Ethiopia commonly dressed
but, they are not Ethiopian cultural cloth classified as unknown. This research overcomes the
needs of experts and everyone who wants to buy and identify capture image and identify
automatically. |
