AI-based plastic waste classification for sorting purposes: A review on recent progresses and challenges

Area of Science:

• Environmental Science and Engineering
• Computer Science and Artificial Intelligence
• Materials Science

Background:

• Rapid plastic waste growth necessitates advanced sorting solutions.
• Artificial Intelligence (AI), including Machine Learning (ML) and Deep Learning (DL), offers automated plastic waste identification and classification.
• Computer vision techniques are crucial for real-time waste stream analysis.

Purpose of the Study:

• To systematically review AI-based methods for plastic waste classification and sorting.
• To critically assess state-of-the-art AI models (CNNs, YOLO, Transformers) and their integration with spectroscopic techniques (NIR, FTIR, Raman).
• To identify challenges and propose future research directions for AI-driven plastic waste management.

Main Methods:

• Systematic literature review following PRISMA guidelines, analyzing 112 articles (2015-2025).
• Analysis of AI classification models (CNNs, YOLO, Transformers) and spectroscopic techniques (NIR, FTIR, Raman).
• Consolidation of performance metrics and mapping of AI model diversity.

Main Results:

• AI models demonstrate high accuracy in plastic waste identification and classification.
• Integration with spectroscopic methods enhances classification precision.
• Key challenges include limited datasets, scalability, and environmental variability.

Conclusions:

• AI offers a promising pathway for efficient and sustainable plastic waste sorting.
• Future research should focus on lightweight models, multi-sensor fusion, and edge-AI for practical deployment.
• This review provides a strategic guide for advancing AI in plastic waste management.