Application of color extraction methods and k-nearest neighbor to determine maturity avocado butter

Abstract

Computerization requires system testing and further system development, namely color feature extraction with KNN. Avocado is one that has a high protein content in it. This research uses the KNN algorithm method and feature extraction in order to get more effective results, the purpose of this research is to make it easier for people to choose the ripeness level of butter avocados because people still don't know about the maturity level of butter avocados. In this study, testing was carried out by bringing the avocado fruit closer to the cellphone camera connected to the researcher's internet, after which the application will automatically match the color of the avocado. to the system, the system will produce output based on that color with output in the form of the ripeness level of the avocado, whether it is ripe, ripe, half ripe, rotten and also generates information on how much longer the avocado will ripen. All stages of system development are carried out by analyzing data first, then taking sample data, training and testing datasets, then the results of the system will become benchmarks. The test data in this study used several types of avocado objects, namely: Raw, Half Ripe, Ripe, Ripe, Rotten. It consisted of 55 data samples consisting of 11 raw avocado samples, 11 half-ripe avocado samples, 11 ripe avocado samples, 11 ripe avocado samples and 11 rotten avocado samples. Obtained euclidean distance values ​​for each type of avocado butter. After that, the sum is done to get the overall level of accuracy by adding up the total euclidean distance with the total euclidean distance for each type of avocado. After getting the added value multiply it by 100%. Then the overall accuracy results obtained are 98.38%.