Zhalgas A., Amirgaliyev B., Boltay B., Shegenova D., Zhylkybay N., Yedilkhan D.
Development of an Intelligent Waste Segregation System Using a Self-Collected Dataset and Deep Learning Methods // Journal of Robotics and Control. — 2026. — Vol. 7, No. 1. — DOI: 10.18196/jrc.v7i1.27247.

Abstract: This paper introduces the design and analysis of an intelligent waste sorting system using deep learning to improve sustainability in waste management. A custom dataset of 2,790 images across six waste categories (plastic, paper, cardboard, glass, metal, and organic waste) was collected, preprocessed, and augmented for model training. Three convolutional neural network architectures—YOLOv8, DenseNet169, and ResNet50v2—were trained and evaluated using accuracy, precision, recall, and F1-score metrics. DenseNet169 achieved the highest overall accuracy (92%), while YOLOv8 demonstrated strong real-time detection performance (91% accuracy), making it suitable for practical deployment in automated waste sorting systems. The results show that transfer learning and fine-tuning significantly improve classification performance, particularly on smaller datasets. The proposed AI-based system demonstrates the effectiveness of computer vision in supporting scalable and sustainable waste management through intelligent, data-driven decision-making.

Link / DOI: https://doi.org/10.18196/jrc.v7i1.27247