Implementation of Lite-RT Based Waste Classification on Android Using Teachable Machine

Muhammad Yusuf, Haidar Khairullah (2026) Implementation of Lite-RT Based Waste Classification on Android Using Teachable Machine. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111942000022-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Pengelolaan sampah merupakan salah satu aspek penting dalam menjaga kebersihan dan kesehatan lingkungan. Salah satu tahapan krusial dalam pengelolaan sampah adalah proses pemilahan antara sampah yang dapat didaur ulang dan yang tidak. Namun, proses pemilahan yang masih dilakukan secara manual cenderung kurang efisien dan berpotensi menimbulkan kesalahan. Penelitian ini bertujuan untuk mengembangkan sebuah aplikasi berbasis Android yang mampu mengklasifikasikan jenis sampah menggunakan pendekatan Edge AI. Model klasifikasi dikembangkan menggunakan platform Google Teachable Machine yang memanfaatkan arsitektur MobileNet, yaitu model Convolutional Neural Network (CNN) ringan yang dirancang untuk perangkat dengan sumber daya terbatas seperti perangkat mobile. Model tersebut kemudian diimplementasikan menggunakan LiteRT sebagai runtime untuk menjalankan model TensorFlow Lite secara on-device. Selain implementasi sistem, penelitian ini juga akan melakukan analisis kinerja model berdasarkan beberapa parameter, seperti accuracy, inference time, dan memory usage, serta mengkaji pengaruh kondisi lingkungan seperti pencahayaan dan jarak objek terhadap hasil prediksi. Hasil penelitian ini diharapkan dapat menunjukkan bahwa penerapan Edge AI pada perangkat mobile mampu memberikan solusi yang cepat, efisien, dan praktis dalam mendukung proses pemilahan sampah secara otomatis.
=========================================================================================================================================
Waste management is an important aspect of maintaining environmental cleanliness and public health. One of the crucial stages in waste management is the classification of waste into recyclable and non-recyclable categories. However, this process is still largely performed manually, which is often inefficient and prone to errors. This research aims to develop an Android-based application capable of classifying types of waste using an Edge AI approach. The classification model is developed using Google Teachable Machine, which utilizes the MobileNet architecture, a lightweight Convolutional Neural Network (CNN) designed for devices with limited computational resources such as mobile devices. The model is then implemented using LiteRT as a runtime to execute the generated TensorFlow Lite model on-device. In addition to system implementation, this research also evaluates the model's performance based on several parameters, including accuracy, inference time, and memory usage. Furthermore, it examines the impact of environmental factors such as lighting conditions and object distance on prediction results. The results of this research are expected to demonstrate that the implementation of Edge AI on mobile devices can provide a fast, efficient, and practical solution for automated waste classification.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Edge AI, Waste Classification, Image Classification, MobileNet, Tensorflow Lite, Android,
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Yusuf Haidar Khairullah
Date Deposited:	31 Jul 2026 03:39
Last Modified:	31 Jul 2026 03:39
URI:	http://repository.its.ac.id/id/eprint/138569

Actions (login required)

View Item