The current system of checking and grading egg quality in the Philippines was done manually one by one using the traditional way where graders exert great effort that resulted in graders' visual stress. To address the problem identified the researchers proposed a scientific way of checking and grading the egg quality by using image processing based non-destructive and cost-effective technique to detect various cracks, dirt, and defect in eggs. Upon testing, the system obtained a total of 91.33% as high-quality eggs and the presence of either crack or dirt while 8.66% were inspected as low quality. For the internal part of each egg, the system achieved 100% detection of the yolk. The main results achieved have been quite promising; the researchers are encouraged to continue the labor of improving the generation of internal and external egg detection.

No abstract

No abstract

Copra in the Philippines is one of the by-products from coconut which contributes as one of the major sources of income of Filipino farmers. During the process of selling the Copra in the market, farmers usually lose in the price competition from the buyers due to the unidentified quality of their Copra. Copra which commonly either overcooked or undercooked are paid as half of the price of the perfectly cooked. This happened due to the lack of information of the farmers in assessing the quality of the processed copra meat. In this study, a Convolutional Neural Network had been evaluated in terms of its accuracy by varying the numbers of convolutional layer filters, the size of filters, and its activation function. The identified best parameters were used to develop a CNN algorithm that classifies the quality of Copra. The algorithm was implemented using Tensorflow in a python environment. A series of tests were applied to the final CNN model. Random images of Copra with identified quality were used as testing data. Out of 120 sample images, the final CNN model performs an overall 86% accuracy. The model was also implemented into a simple android application for validation. The confusion matrix and f-score were used to evaluate the performance of the system.

No abstract