Classification of Javanese characters in mobile devices. In this project, it is a continuation of the learning project cnn digit before,but in this project the introduction of Javanese characters in handwritten form is entered by drawing on the canvas that has been provided, I also use this project as the final project for the pattern recognition course.
In this project, I perform how to use Convolutional Neural networks for pattern recognition and using TensorFlow lite. Starting from Preprocessing the dataset, making the TensorFlow lite model and building the android apps.
- pre-process dataset
- making cnn model
- traning the model
- test the model
- export to tflite
- build first layout
- pre-process from canvas
- test the apps
-
Aksara jawa dataset Before making a model we need a dataset first and I use that from Kaggle Aksara jawa dataset
-
Tensorflow lite An open source deep learning framework for on-device inference
-
Glide Glide is a fast and efficient image loading library for Android
-
Realtime Database a cloud-hosted database provide by google for saving dataset of Javanese script words and reduce the size of the application itself
| we can draw a pattern then classify it and make a sentence from the classification results |  |
| also, we can draw a pattern by following a image |  |