Development of an Intelligent Waste Segregation System Using a Self-Collected Dataset and Deep Learning Methods


Zhalgas A. Amirgaliyev B. Boltay B. Shegenova D. Zhylkybay N. Yedilkhan D.
3 February 2026Department of Agribusiness, Universitas Muhammadiyah Yogyakarta

Journal of Robotics and Control (JRC)
2026#7Issue 13393 - 3404 pp.

This paper introduces the analysis and design of an intelligent waste sorting system with deep learning to analyze waste and increase sustainability in waste management. The dataset (2,790 images in 6 categories of waste, including plastic, paper, cardboard, glass, metal and organic waste) was gathered using custom methods and further preprocessed and augmented to train models. Three convolutional neural network designs, including YOLOv8, DenseNet169, and ResNet50v2, were trained and evaluated in terms of accuracy, precision, and recall and F1-score. DenseNet169 was most accurate (92% in total), and balanced in all evaluation measures, whereas YOLOv8 showed excellent real-time detection with 91% accuracy, which qualifies the latter to be utilized in the practice of waste sorting. The findings substantiate the claims that transfer learning and fine-tuning enhance the reliability of classification mostly on smaller datasets. The suggested AI-based system illustrates the ability of computer vision to adhere to the vision of scalable, efficient, and sustainable waste management through waste management solutions based on the intelligent and data-driven decision-making model.

Artificial Intelligence , Computer Vision , Convolutional Neural Networks , Deep Learning , Waste Management

Text of the article Перейти на текст статьи

Department of Computational and Data Science, Astana IT University, Astana, Kazakhstan
Department of Computer Engineering, Astana IT University, Astana, Kazakhstan

Department of Computational and Data Science
Department of Computer Engineering

10 лет помогаем публиковать статьи Международный издатель

Книга Публикация научной статьи Волощук 2026 Book Publication of a scientific article 2026