Scholarlink

<p>Food quality control is crucial in the agricultural sector, especially in the case of table olives intended for fresh consumption, whereby selection and classification of drupes are important in post-harvest to contain product losses and in pre- and post-processing to attract consumer attention. In recent years, autonomous automatic inspection tools based on advanced technologies, i.e. artificial intelligence (AI)-powered image analysis, have been developed to objectively and non-invasively determine certain optical products characteristics related to quality parameters. This study proposes a real-time quality classification process of freshly harvested table olives belonging to three different Italian cultivars, both in terms of ripening stage and presence/absence of external and internal defects, using a mechano-optoelectronic system equipped with a VIS-NIR hyperspectral camera and an RGB camera, enhanced with AI algorithms. A shallow neural network model was developed for VIS-NIR images to identify cultivars, ripening degree and internal defects of olives, achieving better performance (99% accuracy) for cultivar classification. RGB images were processed with You Only Look Once (YOLO) Convolutional Neural Networks to discriminate regular from damaged olives, both by single cultivar and all cultivars together. The model considering all cultivars together showed good performance in detecting damaged olives, with a mean test accuracy of 0.87. The real-time mechano-optoelectronic system used to analyze olives through RGB and VIS-NIR images, in combination with AI algorithms, proved to be a good tool to qualitatively classify olives, regardless of their maturity and cultivar. Through appropriate scale-up the system could be used and implemented in processing lines for industrial purposes, making the agricultural system more effective and efficient and achieving higher quality production.</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

Artificial intelligence approaches for real-time table olive cultivars quality assessment

  • 期刊论文

  • 发表日期：

    2026年1月21日

查看全文
  • Rossella Manganiello,
  • Lavinia Moscovini,
  • Luciano Ortenzi,
  • Simone Figorilli,
  • Federico Pallottino,
  • Corrado Costa,
  • Simone Vasta,
  • Simona Violino,
  • Francesca Antonucci

摘要

Food quality control is crucial in the agricultural sector, especially in the case of table olives intended for fresh consumption, whereby selection and classification of drupes are important in post-harvest to contain product losses and in pre- and post-processing to attract consumer attention. In recent years, autonomous automatic inspection tools based on advanced technologies, i.e. artificial intelligence (AI)-powered image analysis, have been developed to objectively and non-invasively determine certain optical products characteristics related to quality parameters. This study proposes a real-time quality classification process of freshly harvested table olives belonging to three different Italian cultivars, both in terms of ripening stage and presence/absence of external and internal defects, using a mechano-optoelectronic system equipped with a VIS-NIR hyperspectral camera and an RGB camera, enhanced with AI algorithms. A shallow neural network model was developed for VIS-NIR images to identify cultivars, ripening degree and internal defects of olives, achieving better performance (99% accuracy) for cultivar classification. RGB images were processed with You Only Look Once (YOLO) Convolutional Neural Networks to discriminate regular from damaged olives, both by single cultivar and all cultivars together. The model considering all cultivars together showed good performance in detecting damaged olives, with a mean test accuracy of 0.87. The real-time mechano-optoelectronic system used to analyze olives through RGB and VIS-NIR images, in combination with AI algorithms, proved to be a good tool to qualitatively classify olives, regardless of their maturity and cultivar. Through appropriate scale-up the system could be used and implemented in processing lines for industrial purposes, making the agricultural system more effective and efficient and achieving higher quality production.