Скачать или смотреть GDIID 2025 | ID 5024 Title: RECYCLABLE WASTE CLASSIFICATION USING YOLO-BASED CNN

Project ID: 5024
Title: RECYCLABLE WASTE CLASSIFICATION USING YOLO-BASED CONVOLUTIONAL NEURAL NETWORK (CNN)
Category IID: SCIENCE & TECHNOLOGY

Effective waste sorting is crucial for improving recycling efficiency and mitigating environmental challenges. However, traditional manual sorting methods are time-consuming, labour-intensive, and prone to inaccuracies, leading to contamination in recycling streams. Such contamination decreases the quality and economic value of recyclables, increases processing costs, and hinders sustainable waste management efforts. To address these challenges, this study proposes an automated recyclable waste classification system utilizing a YOLO-based Convolutional Neural Network (CNN) to enhance waste sorting accuracy and efficiency. This study follows the CRISP-DM (Cross-Industry Standard Process for Data Mining) methodology, which provides a structured framework comprising business understanding, data understanding, data preparation, modelling, evaluation, and deployment. The YOLOv8 model was selected due to its superior object detection capabilities, demonstrating over 90% classification accuracy in distinguishing plastic, paper, and glass waste. A dataset sourced from Kaggle was used for model training, consisting of 500 labelled images per category, ensuring robust learning and classification. Preprocessing techniques such as data augmentation, resizing, and normalization were applied to optimize
model performance. In addition to model training, a user-friendly interface was developed to facilitate real-time waste detection, making the system accessible to waste management professionals, municipalities, and recycling facilities. The integration of AI-based waste classification streamlines recycling processes, reduces human effort, and minimizes sorting errors, ultimately leading to higher recycling rates and lower landfill dependency. This study highlights the potential of AI-driven automation in waste management, demonstrating that deep learning-based classification can enhance recycling efficiency, reduce contamination, and contribute to environmental sustainability. By leveraging advanced image recognition techniques and a structured data mining approach, this project offers a scalable and practical solution for waste classification, aligning with global sustainability goals.