Scholarlink

<p>Due to their high perishability, mangoes are susceptible to internal browning during ripening as a result of incorrect handling, storage, and environmental factors. This browning significantly affects post-harvest quality, reducing both marketability and consumer satisfaction. In this research, Near-Infrared Spectroscopy and 3D Deep Neural Networks for Mango Internal Browning Detection (3DSDNN-IBM-NIS) are proposed. The process begins with collecting input images from commercial orchards in Petrolina, Brazil, followed by a pre-processing phase using Broad Collaborative Filtering (BCF) for resizing, noise reduction, and normalization. The pre-processed image is then fed into 3D Shallow Deep Neural Network (3DSDNN) to detect internal browning in mangoes, classifying them as healthy or brown. The Modified Adaptive Ant Colony Optimization Algorithm (MAACOA) is used to optimize the 3DSDNN in order to improve classification accuracy. The proposed 3DSDNN-IBM-NIS method is excluded in Python and evaluated using performance metrics like Accuracy, Recall, Precision, Sample Frequency, and RMSEC (Root Mean Square Error Calibration). Comparing the proposed 3DSDNN-IBM-NIS approach to the current one, it covers 16.42%, 23.36%, and 19.27% greater accuracy, and 16.26%, 34.41%, and 23.26% higher precision, compared with existing methods, Automatic Estimation of Polyphenol Oxidase and Peroxidase Activity in Bell Peppers Using Vis/NIR Spectroscopy (NDE-NIS-SVM), the classification of mango disorder (CMD-RCNN), internal browning in mangoes utilizing visible as well as near-infrared spectroscopy (IBM-NIS-ANN). The 3DSDNN-IBM-NIS method outperforms existing techniques, providing a robust and efficient solution for detecting mango internal browning in post-harvest management.</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 mango internal browning: integrating near-infrared spectroscopy with 3D shallow deep neural networks and adaptive ant colony optimization for accurate quality assessment

  • 期刊论文

  • 发表日期：

    2025年8月7日

查看全文
  • Maddu Srinivasa Rao,
  • S. M. K. M. Abbas Ahmad,
  • Dustakar Surendra Rao,
  • Kummari Ramyasree,
  • P. Sandeep,
  • Maznu Shaik

摘要

Due to their high perishability, mangoes are susceptible to internal browning during ripening as a result of incorrect handling, storage, and environmental factors. This browning significantly affects post-harvest quality, reducing both marketability and consumer satisfaction. In this research, Near-Infrared Spectroscopy and 3D Deep Neural Networks for Mango Internal Browning Detection (3DSDNN-IBM-NIS) are proposed. The process begins with collecting input images from commercial orchards in Petrolina, Brazil, followed by a pre-processing phase using Broad Collaborative Filtering (BCF) for resizing, noise reduction, and normalization. The pre-processed image is then fed into 3D Shallow Deep Neural Network (3DSDNN) to detect internal browning in mangoes, classifying them as healthy or brown. The Modified Adaptive Ant Colony Optimization Algorithm (MAACOA) is used to optimize the 3DSDNN in order to improve classification accuracy. The proposed 3DSDNN-IBM-NIS method is excluded in Python and evaluated using performance metrics like Accuracy, Recall, Precision, Sample Frequency, and RMSEC (Root Mean Square Error Calibration). Comparing the proposed 3DSDNN-IBM-NIS approach to the current one, it covers 16.42%, 23.36%, and 19.27% greater accuracy, and 16.26%, 34.41%, and 23.26% higher precision, compared with existing methods, Automatic Estimation of Polyphenol Oxidase and Peroxidase Activity in Bell Peppers Using Vis/NIR Spectroscopy (NDE-NIS-SVM), the classification of mango disorder (CMD-RCNN), internal browning in mangoes utilizing visible as well as near-infrared spectroscopy (IBM-NIS-ANN). The 3DSDNN-IBM-NIS method outperforms existing techniques, providing a robust and efficient solution for detecting mango internal browning in post-harvest management.