Scholarlink

<p>Fruit quality detection plays a&#xa0;substantial role in safeguarding human health and consumer satisfaction, and in minimizing post-harvest losses. The research proposes a&#xa0;generative adversarial network ensemble activation-enabled bi-directional long short-term memory (GEABTM) model for automated fruit quality detection, which synergizes the accuracy of fruit quality detection. Moreover, the deep histogram pixel flow feature is used, which enables robustness in detecting fruit quality, as well as advancing the model with a&#xa0;comprehensive understanding of fruit textures. The GEABTM model enhances the ability to generalize fruit quality detection by generating diverse training data as well as learning complex patterns, which results in accurate detection. Moreover, the model achieves superior performance in terms of accuracy of 97.99%, F1-score of 97.8%, precision of 98.37%, and recall of 97.25% using a&#xa0;training percentage of&#xa0;90%.</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

GEABTM: A Generative Adversarial Network Ensemble Activation-Enabled Bi-directional Long Short-Term Memory Model for Automated Fruit Quality Detection

  • 期刊论文

  • 发表日期：

    2025年12月4日

查看全文
  • Puja Cholke,
  • Ashutosh Madhukar Kulkarni,
  • Supriya S. Telsang,
  • Vishal S. Thakare,
  • Jayshri S. Sonawane,
  • Rajashree Tukaram Gadhave

摘要

Fruit quality detection plays a substantial role in safeguarding human health and consumer satisfaction, and in minimizing post-harvest losses. The research proposes a generative adversarial network ensemble activation-enabled bi-directional long short-term memory (GEABTM) model for automated fruit quality detection, which synergizes the accuracy of fruit quality detection. Moreover, the deep histogram pixel flow feature is used, which enables robustness in detecting fruit quality, as well as advancing the model with a comprehensive understanding of fruit textures. The GEABTM model enhances the ability to generalize fruit quality detection by generating diverse training data as well as learning complex patterns, which results in accurate detection. Moreover, the model achieves superior performance in terms of accuracy of 97.99%, F1-score of 97.8%, precision of 98.37%, and recall of 97.25% using a training percentage of 90%.