Classification of Child Items in a Gold Tree using Support Vector Machine Classifier

Sorting of images has been a challenge in Machine Learning Algorithms over the years. Various algorithms have been proposed to sort an image but none of them are able to sort the image clearly. The drawback of the existing systems is that the sorted image is not clearly identified. So, to overcome t...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International journal of recent technology and engineering 2019-11, Vol.8 (4), p.3208-3216
Hauptverfasser: Sabeenian, Dr.R.S., Paramasivam, Dr. M.E., R, Anand, S, Mr. Hariharan
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Sorting of images has been a challenge in Machine Learning Algorithms over the years. Various algorithms have been proposed to sort an image but none of them are able to sort the image clearly. The drawback of the existing systems is that the sorted image is not clearly identified. So, to overcome this drawback we have proposed a novel approach to sort the children of a tree and match them with the existing designs. The images will be sorted on the basis of the class of the image. The images are taken from the image and manual binning of those images are done. Then the images are trained and tested. GLCM feature is extracted from the trained and tested images which are later on fed to the SVM classifier. The classification of image is then done with the help of SVM classifier. Around 7000 images are trained on SVM and used for classification. More than 300 different classes have been created in the database for comparison. Real-time images of child items are captured and fed to the SVM for classifying. The main application of this image is the use in distinguishing the designs in the ornaments. The various parts of the ornaments can be differentiated clearly. Thus, the proposed method is precise as compared to the existing methods.
ISSN:2277-3878
2277-3878
DOI:10.35940/ijrte.D8026.118419