Scholarlink

<p><?noindent??>Effective waste classification is a key challenge in modern environmental management and recycling systems. Despite the success of deep learning models in visual recognition, their high computational cost often limits practical deployment. This study aims to develop an efficient and accurate waste classification framework by integrating transfer learning with machine learning techniques. Specifically, a pre-trained VGG16 network is utilized for feature extraction, and a K-Nearest Neighbors classifier is applied for final classification.</p><p><?noindent??>The approach is evaluated on a dataset of high-resolution waste images categorized into nine classes. Experimental results demonstrate an outstanding 99.9% classification accuracy, outperforming several state-of-the-art deep learning and machine learning models. The findings highlight the effectiveness of combining deep feature representations with lightweight classifiers for real-world, automated waste management applications.</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

Real-world waste classification using integrated approach of transfer learning features with machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月20日

查看全文
  • Nimra Akram,
  • Muhammad Shadab Alam Hashmi,
  • Irene Delgado Noya,
  • Helena Garay,
  • Imran Ashraf

摘要

Effective waste classification is a key challenge in modern environmental management and recycling systems. Despite the success of deep learning models in visual recognition, their high computational cost often limits practical deployment. This study aims to develop an efficient and accurate waste classification framework by integrating transfer learning with machine learning techniques. Specifically, a pre-trained VGG16 network is utilized for feature extraction, and a K-Nearest Neighbors classifier is applied for final classification.

The approach is evaluated on a dataset of high-resolution waste images categorized into nine classes. Experimental results demonstrate an outstanding 99.9% classification accuracy, outperforming several state-of-the-art deep learning and machine learning models. The findings highlight the effectiveness of combining deep feature representations with lightweight classifiers for real-world, automated waste management applications.