Attention-Enhanced YOLOv8-Based Plastic ResinClassification for Intelligent Recycling Support

This study presented an enhanced YOLOv8-based frame-work for image-level identification of seven plastic resin categories: HDPE, LDPE, PET, PP, PS, PVC, and Other. The method combines a YOLOv8s-cls backbone with trans-fer learning, focal loss, label smoothing, MixUp, CutMix, Softmax-based confidence estimation, and Grad-CAM visu-alization. Relative to the reported 79.05% accuracy of the lightweight YOLOv8n-cls baseline, the enhanced model achieved a re-ported accuracy of 88.00%. The model provides not only the predicted resin class but also confidence information, ranked alternatives, and a visual explanation, making it more suitable for operator-assisted recycling applications than a classification-only pipeline. Although RGB-based recognition cannot completely resolve polymers with extremely similar appearances, the findings demonstrate the potential of deep-learning-assisted resin clas-sification for low-cost recycling applications. Future develop-ment should focus on larger real-world datasets, multimodal sensing, resin-code recognition, edge deployment, and integra-tion with automated sorting systems.

[1]     S. Serranti and G. Bonifazi, “Techniques for separation of plastic wastes,” in Use of Recycled Plastics in Eco-Efficient Concrete. Wood-head Publishing, 2019, pp. 9–37. [2]     M. Rani, N. Marchesi, L. Federici, C. Rovelli, I. Alessandri, F. Vassalini,G. Ducoli, L. Borgese, and A. Zacco, “Miniaturized near-infrared spectrometer in plastic waste sorting,” Materials, vol. 12, no. 17, p. 2740, 2019. [3]     Q. Duan and J. Li, “Classification of common household plastic wastes combining multiple methods based on near-infrared spectroscopy,” ACS ES&T Engineering, vol. 1, no. 7, pp. 1065–1073, 2021. [4]     T. Sousa, A. Matos, and J. Oliveira, “Deep learning for plastic waste classification,” in Proc. Int. Conf. Intelligent Systems and Data Science, 2022. [5]     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 Management, vol. 138, pp. 274–284, 2022. [6]     S. Singh, S. Hait, and A. Thakur, “Hyperspectral imaging-based classi-fication of post-consumer thermoplastics,” Resources, Conservation and Recycling, 2023. [7]     J. Choi, J. Lim, and S. Yoo, “Advancing plastic waste classification and recycling efficiency: Integrating image sensors and deep learning algorithms,” Applied Sciences, vol. 13, no. 18, p. 10224, 2023. [8]     S. Kunwar, B. R. Owabumoye, and A. S. Alade, “Plastic waste classi-fication using deep learning: Insights from the WaDaBa dataset,” arXiv preprint arXiv:2412.20232, 2024. [9]     A. Arishi, “Real-time household waste detection and classification using YOLOv8 deep learning model,” Applied Sciences, 2025. [10]    J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition, 2016, pp. 779–788.