<![CDATA[This research aims to design and implement a cloud-based backend system for waste classification, utilizing services provided by Google Cloud Platform (GCP). The development focuses on using Cloud Run as a serverless platform for running backend APIs, AutoML Vision as a machine learning-based image classification model, and Cloud Firestore as a NoSQL database to store classification results. The system is backend-only, designed to receive image input, process it through a trained classification model, and automatically store the results in the database. The methodology employed in this study is Research and Development (R&D) with a quantitative data analysis approach. System testing was conducted using black-box testing and Postman API to verify that the system functions as intended. The evaluation involved measuring the classification accuracy of the AutoML Vision model, API response times, and data storage reliability in Firestore. The results show that the model successfully classified two waste categories (plastic and paper) with an accuracy of 90%, an average API response time of 1033 milliseconds, and consistent data storage without loss. The findings indicate that the developed backend system operates optimally and aligns with the research objectives. This system has the potential to be further developed into a cloud-based technological solution to support efficient and automated waste management.]]>
Copyrights © 2026
