Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2024
- Goals
- Record
- 2a1f2371-d99f-4afb-b316-a1d073ac67ca
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted an activity developing an image-based litter classification model using a Deep Learning DenseNet architecture. The activity addressed the growing waste management problem and the need for innovative solutions to support efficiency in sorting and recycling. The activity participants were Zulhusni, Munis; Sari, Christy Atika; and Rachmawanto, Eko Hari.
The model was designed to address the challenge of accurate waste classification, which is affected by variability in visual characteristics between waste types. The activity used diverse training datasets and classified waste into ten different categories. The reported results showed that the model achieved an overall accuracy rate of 93%. The model also showed an excellent ability to identify and classify specific materials, including batteries, biological materials, and brown glass. The source identified challenges in metal and plastic classification.
The activity relates to waste sorting and recycling through the development of an image-based classification model. The source describes the potential of Deep Learning technology in waste management systems to improve sorting processes and increase recycling efficiency. As an institutional activity of Universitas Dian Nuswantoro, this work documents the development and reported classification performance of a DenseNet-based model, while the source-supported findings remain focused on the stated categories, materials, accuracy, and classification challenges.