GarbageIQ: A Real-Time AI Powered Waste Classification and Recycling Guidance System Using YOLOv8

This paper has presented GarbageIQ, a real-time AI-powered waste classification and recycling guidance system that integrates YOLOv8 object detection with a structured waste ontology and an interactive Streamlit dashboard. The system achieves a [email protected] of 89.3 percent at 38.4 frames per second on commodity GPU hardware, outperforming YOLOv7 by 6.1 mAP points and Faster R-CNN by 18.4 mAP points, while operating with an end-to-end inference latency of 23 ms—well within the real-time interaction threshold. The three-tier edge architecture (acquisition, inference, presentation) ensures offline operability, data privacy, and modular extensibility. The deterministic COCO-to-Waste ontology covering 44 object classes across nine waste categories provides actionable recycling instructions without requiring domain-specific training data, making the system immediately deployable on existing COCO-pretrained weights. GarbageIQ demonstrates that the combination of state-of-the-art single-stage detection and lightweight rule-based ontological mapping can deliver practical waste intelligence at the edge, lowering the barrier for municipalities, educational institutions, and household users to engage with correct disposal behaviour. Future work will focus on fine-tuning the detection backbone on purpose-built waste datasets, integrating     geolocation-aware           facility recommendation APIs, and extending guidance to 20+ additional languages, positioning GarbageIQ as a scalable foundation for intelligent waste management at city and national scale.

[1]    World Bank, 'What a Waste 2.0: A Global Snapshot of Solid Waste Management to 2050,' The World Bank, Washington D.C., 2018. [Online]. Available: https://datatopics.worldbank.org/what-a-waste/ [2]    K. M. Thung and T. H. Chan, "A Survey of Image Quality Measures," in Proc. Int. Conf. Technical Postgraduates (TECHPOS), Kuala Lumpur, 2009, pp. 1–4. [3]    A. E. Aral, R. B. Keskin, and U. Yildirim, "Classification of TrashNet Dataset Based on Deep Learning Models," in Proc. IEEE Int. Conf. Big Data (BigData), Seattle, WA, 2018,pp.  2058–2062,      doi: 10.1109/BigData.2018.8622212. [4]    A. V. Vo, M. Q. Tran, and N. D. Nguyen, "Waste Classification Using Deep Learning on Mobile Devices," J. Sci. Technol. Issues, vol. 9, no. 2, pp. 34–41, 2021. [5]    S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Trans. Pattern Anal. Mach. Intell., vol. 39, no. 6, pp. 1137–1149,      Jun.   2017,           doi: 10.1109/TPAMI.2016.2577031. [6]    S. Majchrowska, A. Mikołajczyk, M. Ferlin,Z. Klawikowska, M. A. Plantykow, A. Kwasigroch, and K. Majek, "Deep Learning-Based Waste Detection in Natural and Urban Environments," Waste Manag., vol. 138,  pp.  274–284,  Feb.  2022,  doi:10.1016/j.wasman.2021.12.001. [7]    J. Bobulski and M. Kubanek, "Deep Learning for Plastic Waste Classification System," Appl. Comput. Intell. Soft Comput., vol. 2021, Art.no.      6626948,   2021,     doi: 10.1155/2021/6626948. [8]    J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Las Vegas, NV, 2016, pp. 779–788, doi: 10.1109/CVPR.2016.91. [9]    C.-Y. Wang, A. Bochkovskiy, and H.-Y. M. Liao, "YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Vancouver, BC, 2023, pp. 7464–7475, doi: 10.1109/CVPR52729.2023.00721. [10] Y. Li, B. Chen, and J. Liu, "Enhanced YOLOv7 with Deformable Convolution for Waste Package Detection," Expert Syst. Appl., vol. 213, Art. no. 118975, Mar. 2023, doi: 10.1016/j.eswa.2022.118975.