Authors:
R. Regin, M. Arjun Raj, M. A. Thinesh, S. S. Mukhil Varmann, M. Mohamed Thariq, Mykhailo Paslavskyi

Addresses:
Department of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India. Department of Computer Science and Engineering, Dhaanish Ahmed College of Engineering, Chennai, Tamil Nadu, India. Department of Computer Science, National Forestry University of Ukraine, Lviv, Lviv Oblast, Ukraine.

Abstract:

Waste generation attributable to urban life, changed consumer behaviour, and increased use of disposable lifestyles, among other contributing factors, have emerged as a major environmental and public health challenge. Improper waste management leads to pollution of land, air, and water, thereby endangering both ecosystems and communities. Slow, error-prone, and less efficient, Lesly's slower speed is incapable of achieving efficient results in traditional manual sorting methods. To address this, we propose a novel AI-based solution in this paper: a YOLOv8-based garbage detection model enhanced with a GhostNet head to further improve speed and accuracy. YOLOv8 is well-known for its real-time object detection capabilities, while GhostNet offers a lightweight yet powerful architecture that reduces processing load without compromising features. Trained on a wide variety of waste-detection datasets and complemented with rich data augmentation, this system is designed for real-world environmental conditions. It achieves impressive precision (91%), recall (88%), F1-Score (89%), and mAP@50 (0.936), proving it is well-suited for adoption in smart waste management systems. This would not only provide higher accuracy in sorting recyclables but also result in less landfilling, resource conservation, and more sustainable practices. Our research sheds light on how deep learning can be transformative in addressing waste-related issues and points toward the future, when technology will help create a cleaner, healthier planet.

Keywords: Waste Management; Garbage Detection; Real-Time Object; Urban Life; Deep Learning; Object Detection; AI in Waste Management; Data Augmentation.

Received on: 01/09/2024, Revised on: 15/11/2024, Accepted on: 20/12/2024, Published on: 05/06/2025

DOI: 10.69888/FTSCS.2025.000432

FMDB Transactions on Sustainable Computing Systems, 2025 Vol. 3 No. 2, Pages: 68-83

• Views : 98
• Downloads : 5

Download PDF