Determination of Shelled Corn Damages using Colored Image Edge Detection with Convolutional Neural Network
Analyn N. Yumang, Glenn V. Magwili, Sev Kyle C. Montoya, Corleone Jorel G. Zaldarriaga
Abstract
In the Philippines corn is one of the top agricultural products produced in the country, specifically yellow corn. It is distributed in various cities and provinces to consumers. It is important that the corn kernels to undergo quality assurance before releasing them to the consumers. The methods for evaluating and qualifying corn kernels that are employed by most farms in the country are only done by manual human inspection and these methods are inconsistent which results to inaccurate findings. This is more prevalent when dealing with large amounts of kernels that need to be qualified. This study offers to reduce those inconsistencies by implementing a neural network-assisted method of inspection. The damages to corn kernels can be determined by its physical attributes and as such, the neural network will easily detect the type of damage within a given sample. Aside from the healthy kernels, the types of damage that was included in this study are the following: drier damage, heat damage, heat damage (drier phase), OCOL (Other Color) Type A and OCOL Type B. The neural network that will be used will be a Convolutional Neural Network wherein the images of the samples are subjected to layers of processing. This study also uses Colored Image Edge Detection. The detection method used in this study has obtained an accuracy rating of 96.66%.