Scholarlink

<p>White fig is a valuable fruit crop known for their nutritional benefits significance; however, determining ripeness remains a challenge due to reliance on subjective visual assessment. In this study, a machine vision-based method was developed for the non-destructive classification of white fig ripeness. Samples representing three ripeness stages (unripe, ripe, and overripe) were imaged using a high-resolution imaging setup, producing a dataset of 4,500 augmented images. Three convolutional neural network (CNN) architectures—ResNet-50, VGG16, and MobileNetV2—were trained and evaluated for classification performance. The highest accuracy was achieved by ResNet-50, with training and test accuracies of 99.82% and 94.07%, respectively, surpassing the test accuracies of VGG16 and MobileNetV2, which were 93.37% and 88.07%, respectively. These results demonstrate the effectiveness of deep learning models for real-time, automated detection of white fig ripeness, providing a reliable alternative to traditional manual methods. This approach shows strong potential for application in precision agriculture and post-harvest quality control systems.</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

Detection of white fig ripeness stages using deep learning models

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • Ehsan Sheidaee,
  • Pourya Bazyar

摘要

White fig is a valuable fruit crop known for their nutritional benefits significance; however, determining ripeness remains a challenge due to reliance on subjective visual assessment. In this study, a machine vision-based method was developed for the non-destructive classification of white fig ripeness. Samples representing three ripeness stages (unripe, ripe, and overripe) were imaged using a high-resolution imaging setup, producing a dataset of 4,500 augmented images. Three convolutional neural network (CNN) architectures—ResNet-50, VGG16, and MobileNetV2—were trained and evaluated for classification performance. The highest accuracy was achieved by ResNet-50, with training and test accuracies of 99.82% and 94.07%, respectively, surpassing the test accuracies of VGG16 and MobileNetV2, which were 93.37% and 88.07%, respectively. These results demonstrate the effectiveness of deep learning models for real-time, automated detection of white fig ripeness, providing a reliable alternative to traditional manual methods. This approach shows strong potential for application in precision agriculture and post-harvest quality control systems.