Scholarlink

<p>In the ever-evolving field of agriculture and food production, the objective of achieving higher efficiency, accuracy, and sustainability has resulted in a fundamental change in how fruit quality is evaluated. As the need for high-quality product develops and supply networks have become increasingly complex, the incorporation of new technologies, notably machine intelligence, stands out as a revolutionary force. However still at large, most of these require great expertise in domain. This research study, embarks on an exploration of usage of simple images captured by hand held cameras/mobiles and use of pre-trained vision transformers and machine learning in the context of Banana ripening stage detection by classification. The proposed work minimizes the expert knowledge that need to be available at the any stage of supply chain and also proves to be an reliable non invasive method of fruit quality detection by performing with an accuracy of 90.9%. Thus, our proposed light weight model outperforms CNN with fewer training samples and less computation power and can be deployed in simple mobile cameras at any place at any time proving to be easy to use in the fruit supply chain.</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

Non-invasive image based light weight model for banana ripeness stages prediction utilizing machine learning and vision transformers

  • 期刊论文

  • 发表日期：

    2025年7月22日

查看全文
  • S. T. Veena,
  • N. Deepak Somu,
  • J. Pharaneeshwaran

摘要

In the ever-evolving field of agriculture and food production, the objective of achieving higher efficiency, accuracy, and sustainability has resulted in a fundamental change in how fruit quality is evaluated. As the need for high-quality product develops and supply networks have become increasingly complex, the incorporation of new technologies, notably machine intelligence, stands out as a revolutionary force. However still at large, most of these require great expertise in domain. This research study, embarks on an exploration of usage of simple images captured by hand held cameras/mobiles and use of pre-trained vision transformers and machine learning in the context of Banana ripening stage detection by classification. The proposed work minimizes the expert knowledge that need to be available at the any stage of supply chain and also proves to be an reliable non invasive method of fruit quality detection by performing with an accuracy of 90.9%. Thus, our proposed light weight model outperforms CNN with fewer training samples and less computation power and can be deployed in simple mobile cameras at any place at any time proving to be easy to use in the fruit supply chain.