Scholarlink

<p>Mango grading is a vital quality-checking process in the agriculture and food industry. Manual quality assessment is a time-consuming and cost-effective process. Automatic Mongo grade detection is possible in this decade with the help of machine learning, computer vision and image processing techniques. The work proposed a novel adaptive system for mango grade detection. It measures the physical parameters of the mangoes such as size, length, width and defective areas from images. From the physical parameters, feature vectors (FV) of mango are framed with hugeness (H), fineness (F) and roundness (R). An adaptive feature vector (AFV) is generated from FV using grade rule and default thresholding. In real-time, a graphical user interface (GUI) helps the user to change the grading criteria with adaptive thresholding. This adaptability makes the grading process easy for end users such as farmers, vendors, consumers, exporters, and fruit processors to scan and grade the quality of mangoes quickly. In the end, AFV is used for grade detection by deploying six machine learning (ML) classifiers like Decision Tree, Support Vector Machine, K-Nearest neighbour, Naive Bayes, Random Forest, and Logistic Regression. Ensemble learning is employed for final grading from ML classifiers using the maximum voting method. The proposed system is trained and tested with five existing datasets and detects a grade accuracy of 80.26% using default thresholding. In addition, the proposed system is experiments with a real-time dataset in which 92% of the mangoes are accurately using adaptive grade rule.</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

An Automatic Mango Quality Grading System in Smart Agriculture using Novel Adaptive Feature Vector and Ensemble Learning

  • 期刊论文

  • 发表日期：

    2025年2月20日

查看全文
  • Kalaiselvi T.,
  • Veerakumar P.,
  • Thahira Banu Azeez,
  • Somasundaram K.,
  • Praveenkumar S.,
  • Sriramakrishnan P.

摘要

Mango grading is a vital quality-checking process in the agriculture and food industry. Manual quality assessment is a time-consuming and cost-effective process. Automatic Mongo grade detection is possible in this decade with the help of machine learning, computer vision and image processing techniques. The work proposed a novel adaptive system for mango grade detection. It measures the physical parameters of the mangoes such as size, length, width and defective areas from images. From the physical parameters, feature vectors (FV) of mango are framed with hugeness (H), fineness (F) and roundness (R). An adaptive feature vector (AFV) is generated from FV using grade rule and default thresholding. In real-time, a graphical user interface (GUI) helps the user to change the grading criteria with adaptive thresholding. This adaptability makes the grading process easy for end users such as farmers, vendors, consumers, exporters, and fruit processors to scan and grade the quality of mangoes quickly. In the end, AFV is used for grade detection by deploying six machine learning (ML) classifiers like Decision Tree, Support Vector Machine, K-Nearest neighbour, Naive Bayes, Random Forest, and Logistic Regression. Ensemble learning is employed for final grading from ML classifiers using the maximum voting method. The proposed system is trained and tested with five existing datasets and detects a grade accuracy of 80.26% using default thresholding. In addition, the proposed system is experiments with a real-time dataset in which 92% of the mangoes are accurately using adaptive grade rule.