Image-Based Banana Detection Application using K-Nearest Neighbor

Abstract

Banana is a fruit that is very beneficial for the human body, because it contains a lot of nutrients needed by the human body. One serving of banana contains about 110 calories, 30kg carbohydrates and 1 gram of protein. At harvest time, farmers will sort the bananas according to their type, then they will be sorted according to quality (for example, level of maturity, type of banana, suitable for consumption). The sorting process is still done manually by banana farmers in the Lumajang area. Therefore, the results of the sorting process will be less accurate because there are differences in opinion on the quality value of each farmer. This study aims to provide standard values for Lumajang banana farmers. So that each banana that has been harvested by the farmer can be grouped according to predetermined standard values. In this study using 5 color features. The results of the feature extraction process will be processed again using the k-nearest neighboor method. The results of this study are an application that can identify the type of banana and the ripeness of the banana. The identification of banana types and the ripeness of bananas were obtained from the extraction process of color features and continued with the k-nearest neighboor method. The results of accuracy using the K-NN method are 68% accurate K3 and 32% inaccurate.