Machine-Learning

This project is a Product Capstone submission from the C22-PS012 group for Studi Independen Kampus Merdeka of Bangkit 2022. Contributors of this project are:

1. Muhammad Gibran Fadilla (M2002G0076) - Machine Learning
2. Muhammad Faisal Anshory (M2002G0077) - Machine Learning
3. Dzalfikri Ali Zidan (A2009G0963) - Mobile Development
4. M Yayang Setiawan ( A2009G0973) - Mobile Development
5. Taopik Hidayat (C2222W2028) - Cloud Computing
6. Nur Ayu Sulistiani (C7224X2043) - Cloud Computing

Project Background

Culture is an invaluable heritage of our ancestors. Culture is also a national identity that makes us have different characteristics from other countries. One of Indonesia's cultural heritages that are starting to fade and must be preserved is wayang. As the next generation, we are obliged to introduce wayang to future generations and also to the international community. In this modern era, one effective way is to create wayang recognition technology. We take this opportunity as an effort to help preserve Indonesian culture which will be applied with technology. A technology that we named “Dalang” will help us to get to know wayang more deeply starting from its type, area of ​​origin, description, and other information by taking pictures of wayang then "Dalang" will perform classification and display the results of the classification of puppets.

Dataset Collection

We collected the data from Google Image and Bing Image. All images are owned by the respected party, we collected the images only for learning purposes. Our goal is to collect the good and diverse Wayang images for each class (6 classes). The classes consist of:

1. Wayang Beber
2. Wayang Gedog
3. Wayang Golek
4. Wayang Kulit
5. Wayang Suluh
6. Wayang Krucil

Documentation

• Download dataset
• Install libraries
• Create a model
• Train the model
• Validate the model metric
• Save the model
• Convert model to tflite

Acknowledgements

We wouldn't be here without the help of others. If you owe any attributions or thanks, include them here along with any citations of past research.