Identification of defects and assorting in tomatoes using computer vision and machine learning

  • Mr. S. Ashok Kumar, I. Kavipriya, S. Pavithra
Keywords: Convolutional Neural network, DCNN, Faster RCNN, Mask RCNN, Back propagation

Abstract

During the cultivation of crops and in post-harvest process, when the crops are produced in larger scale, there is a need to identify its defects and to grade them in order to increase the earnings. The quality associated with agricultural products is usually often related to their colour. In this study, object recognition models and Diverse deep Convolutional Nerve organs Networks (DCNNs) are combined to recognize the assortment of tomato diseases and to locate the diseased spot in tomato. Many automated colour grading methods generally determine the colouring quality of these types by directly evaluating the coloring of product towards a predefined set associated with reference, which are in three dimensional colour spaces. In this study, a cost efficient, compact and user-friendly tomato grading method is used which is well suitable for commercial production. This grading method uses customer defined database of tomatoes to create categories and selection of colour of interest specific to an application which figures unique set features at the Coefficients. The Red-Green-Blue colour space will be in 3D by default which converts it into a tiny group of colour indices based on application. At first, a new dataset of tomato vegetables are created. In addition to different parameters of those tomatoes, Red and Green color spaces are used to educate the neural community. Once neural community is trained, other tomatoes are split up into different groups according to above skilled network. Thus, this system can be used in real time for tomato grading.

Published
2021-09-17
How to Cite
S. Pavithra, M. S. A. K. I. K. (2021). Identification of defects and assorting in tomatoes using computer vision and machine learning. Design Engineering, 12344-12356. Retrieved from http://www.thedesignengineering.com/index.php/DE/article/view/4395
Section
Articles