Pengenalan Pola Tulang Daun Dengan Jaringan Syaraf Tiruan Backpropagation

  • Alvin Hanjaya Tandrian Universitas Multimedia Nusantara
  • Adhi Kusnadi Universitas Multimedia Nusantara

Abstract

The development of technology has affected many areas of life. Progress in the field of Computer Science can reach other aspect of science. This research apply the knowledge of Computer Science in Biological Science, the one is the morphology of leaf venation. Leaf venation is an important aspect in the process of identification. Therefore, in this research developed the
system that classify the type of leaf venation. This application is used as means of research on the performance of pattern recognition on backpropagation neural network. The system designed using the Java programming and socket programming to transfer data from the mobile device into the computer. Data testing is implemented using Android to facilitate process of
taking the picture. While in the process of training data for the optimal weight applied directly on the server computer by using Java Eclipse. In the stage of image processing is implemented by using the library of Canny edge detection. Data consisted of five categories of leaf vein pattern, with a sample of three leaves for each pattern. Training data using two of the three leaves for each pattern, with 10 images each leaf so that there are 20 images for each pattern, with a total of 100 images for all patterns. Data testing use 10 images from the third leaf to count the accuracy. The system managed to get
the best accuracy by using an image size of 200 x 200 with 100 hidden node with the average accuracy of 76%.

Index Terms— Android, Canny Edge Detection, Java, Neural Networks Backpropagation, Socket Programming

Downloads

Download data is not yet available.
Published
2019-03-19
How to Cite
Tandrian, A., & Kusnadi, A. (2019). Pengenalan Pola Tulang Daun Dengan Jaringan Syaraf Tiruan Backpropagation. Ultima Computing : Jurnal Sistem Komputer, 10(2), 53-58. https://doi.org/https://doi.org/10.31937/sk.v10i2.1063