Scholarlink

<p>This study provides a comprehensive comparative analysis of three state-of-the-art deep learning models—YOLOv8, EfficientDet, and Detectron 2—focusing on their application in waste segregation to address the escalating global waste management challenge. The integration of artificial intelligence (AI) with environmental sustainability has become increasingly critical in this domain. Prior research has demonstrated significant reductions in contamination levels through AI-based sorting systems, yet challenges persist, particularly in applications requiring real-time interpretation of complex visual data, prompting the exploration of advanced AI models. The evolution of waste management through AI technologies, including computer vision, the Internet of Things (IoT), and deep learning, highlights the transformative impact on resource recovery. Traditional machine learning algorithms have faced limitations in fine-grained waste material identification, necessitating the adoption of more sophisticated AI models.The deployment of deep learning models such as YOLOv8, EfficientDet, and Detectron 2 represents a paradigm shift in waste segregation. YOLOv8 stands out due to its architectural advancements, speed, accuracy, and scalability. EfficientDet's ability to adapt to diverse detection requirements through compound scaling enhances its efficiency. Detectron 2, developed by Facebook AI Research (FAIR), is notable for its versatility, accessibility, and extensive applications in object detection. Empirical data highlight the substantial impact of AI on waste management, with studies indicating a 30% reduction in contamination levels. In this study, YOLOv8x achieved a remarkable mAP50 of 42.6%, followed closely by YOLOv8l, YOLOv8m, and YOLOv8s. EfficientDet also delivered exceptional performance in terms of efficiency and precision. Validation data confirm the models' generalization capabilities, while confusion matrices provide insights into specific challenges in waste item classification.This paper addresses the limitations of conventional machine learning models and advocates for the adoption of advanced AI models in waste management. The comparative evaluation underscores the advantages of YOLOv8, EfficientDet, and Detectron 2 in terms of performance, efficiency, and flexibility, establishing them as valuable tools in effective waste segregation. The findings contribute to a nuanced understanding of the strengths and limitations of these models, paving the way for their implementation in real-world waste management scenarios.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A comparative analysis of deep learning models for waste segregation: YOLOv8, EfficientDet, and Detectron 2

  • 期刊论文

  • 发表日期：

    2025年1月29日

查看全文
  • Rohith Sai Midigudla,
  • Tribhangin Dichpally,
  • Uday Vallabhaneni,
  • Yatish Wutla,
  • Divya Meena Sundaram,
  • Sheela Jayachandran

摘要

This study provides a comprehensive comparative analysis of three state-of-the-art deep learning models—YOLOv8, EfficientDet, and Detectron 2—focusing on their application in waste segregation to address the escalating global waste management challenge. The integration of artificial intelligence (AI) with environmental sustainability has become increasingly critical in this domain. Prior research has demonstrated significant reductions in contamination levels through AI-based sorting systems, yet challenges persist, particularly in applications requiring real-time interpretation of complex visual data, prompting the exploration of advanced AI models. The evolution of waste management through AI technologies, including computer vision, the Internet of Things (IoT), and deep learning, highlights the transformative impact on resource recovery. Traditional machine learning algorithms have faced limitations in fine-grained waste material identification, necessitating the adoption of more sophisticated AI models.The deployment of deep learning models such as YOLOv8, EfficientDet, and Detectron 2 represents a paradigm shift in waste segregation. YOLOv8 stands out due to its architectural advancements, speed, accuracy, and scalability. EfficientDet's ability to adapt to diverse detection requirements through compound scaling enhances its efficiency. Detectron 2, developed by Facebook AI Research (FAIR), is notable for its versatility, accessibility, and extensive applications in object detection. Empirical data highlight the substantial impact of AI on waste management, with studies indicating a 30% reduction in contamination levels. In this study, YOLOv8x achieved a remarkable mAP50 of 42.6%, followed closely by YOLOv8l, YOLOv8m, and YOLOv8s. EfficientDet also delivered exceptional performance in terms of efficiency and precision. Validation data confirm the models' generalization capabilities, while confusion matrices provide insights into specific challenges in waste item classification.This paper addresses the limitations of conventional machine learning models and advocates for the adoption of advanced AI models in waste management. The comparative evaluation underscores the advantages of YOLOv8, EfficientDet, and Detectron 2 in terms of performance, efficiency, and flexibility, establishing them as valuable tools in effective waste segregation. The findings contribute to a nuanced understanding of the strengths and limitations of these models, paving the way for their implementation in real-world waste management scenarios.