Open AccessMask R-CNN for Indonesian Shadow Puppet Recognition and Classification
Author(s) -
Ida Bagus Kresna Sudiatmika,
Made Artana,
Nengah Widya Utami,
Made Adi Paramartha Putra,
Eka Grana Aristyana Dewi
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1783/1/012032
Subject(s) - shadow (psychology) , indonesian , convolutional neural network , computer science , artificial intelligence , set (abstract data type) , test set , pattern recognition (psychology) , training set , object (grammar) , computer vision , psychology , linguistics , philosophy , psychotherapist , programming language
Region-based Convolutional Neural Networks (R-CNN) have achieved remarkable achievements in object recognition and detection. In this paper, Mask R-CNN is introduced to detect and recognize Indonesian Shadow Puppet patterns. Indonesian Shadow Puppet (Wayang) is one of the traditional Indonesian arts that depict stories, one of the stories is the Mahabharata. Our aim for conducting this research is to protect Indonesian Puppets (Wayang) from extinction through systemic pattern recognition. The algorithm used in this study is Mask R-CNN. Mask R-CNN training to recognize Indonesian Shadow Puppet requires many labels that are difficult to obtain. Therefore, the authors carried out data augmentation to add datasets to improve training. Furthermore, testing is done using several learning rates. This test is done using a Cloud GPU. The results of the tests show that the accuracy of the training obtained is 92.04% and the accuracy of validation with the learning rate value set is 0.0001. Testing the model in this study successfully tested the puppet pattern.