the Comparative Analysis and Optimization of CNN and MobileNetV2 Using Data Augmentation and Fine-Tuning for Waste Classification
DOI:
https://doi.org/10.31937/ti.v18i1.4687Abstract
Waste management has become an increasingly critical environmental issue that requires effective technology-based solutions. One promising approach is the application of deep learning methods for automatic waste classification. This study aims to compare the performance of Convolutional Neural Network (CNN) and MobileNetV2 models, as well as to evaluate the impact of data augmentation and optimization techniques on model performance.
The dataset consists of 2,527 waste images categorized into six classes: plastic, paper, glass, metal, cardboard, and trash. The training process incorporates data augmentation, transfer learning, and optimization techniques such as dropout, learning rate adjustment, and early stopping. Experimental results show that the CNN model achieves a validation accuracy of 48.5%, whereas MobileNetV2 attains 69.8%.
Furthermore, MobileNetV2 demonstrates better generalization performance, indicated by a smaller gap between training and validation accuracy. These findings suggest that MobileNetV2 combined with appropriate optimization techniques is more effective than CNN for waste classification tasks.
Keywords— CNN, Data Augmentation, MobileNetV2, Transfer Learning, Waste Classification
Downloads
Additional Files
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Deas Aghellar, Aswan Supriyadi Sunge
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution-ShareAlike International License (CC-BY-SA 4.0) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
Copyright without Restrictions
The journal allows the author(s) to hold the copyright without restrictions and will retain publishing rights without restrictions.
The submitted papers are assumed to contain no proprietary material unprotected by patent or patent application; responsibility for technical content and for protection of proprietary material rests solely with the author(s) and their organizations and is not the responsibility of the ULTIMATICS or its Editorial Staff. The main (first/corresponding) author is responsible for ensuring that the article has been seen and approved by all the other authors. It is the responsibility of the author to obtain all necessary copyright release permissions for the use of any copyrighted materials in the manuscript prior to the submission.
