Tomato grading system with computer vision

Tee, Say Jin (2020) Tomato grading system with computer vision. [Project Paper] (Submitted)

Text FK 2020 29.pdf Download (1MB)

Abstract

In Malaysia, tomato is one of the few vegetables that are able to penetrate domestic retail sector as well as export. Grading is a key measure to commercialize the produce where surface defects and size are two important parameters to be checked in order to determine its quality. For postharvest treatments of tomato, surface defects detection and size measuring are carried out independently. This has greatly increased the labour usage, energy and cost. Moreover, the difficulty with the traditional machine learning approach makes surface defects inspection a tedious work. This is because it is necessary to choose which features are important in each given image. As the number of classes to classify increases, feature extraction becomes more and more cumbersome. It is also up to the Computer vision (CV) engineer‟s judgment and a long trial and error process to decide which features best describe different classes of objects. With the great progress of deep learning in the field of computer vision, the application of deep learning in image classification has become more popular. The proposed tomato classification system consists of two main parts: 1) Surface defect detection and 2) Size measuring. For the first part of the model, deep learning approach, specifically Convolutional Neural Network (CNN) is applied as it involves non-linear and complex parameters. For the second part of the model, an algorithm is developed with OpenCV library for tomato size measuring by calculating Euclidean Distance. Both of these models are merged as a robust classifer. Both of these models are developed and tested separately. The accuracy obtained for surface defect detection model developed in this study is 97.61% while the algorithm of tomato size measuring give the root-mean-square-error (RMSE), normalized RMSE, Index of agreement and Nash-Sutcliffe Model Efficiency (NSE) values of 0.258 mm, 0.00517, 1.000 and 0.999, respectively

Item Type:	Project Paper
Faculty:	Faculty of Engineering
Depositing User:	Ms. AZLINA ZAINAL ABIDIN
Date Deposited:	15 Nov 2022 02:59
Last Modified:	15 Nov 2022 02:59
URI:	http://psaspb.upm.edu.my/id/eprint/619

Actions (login required)

View Item