Klasifikasi Jenis Buah Kelengkeng Dengan Metode K-Nearest Neighbor (KNN) Berdasarkan Citra Warna Buah

Abstract

In the study titled "Classification of Longan Fruit Types Using KNN Method Based on Fruit Color Images" with the use of the TensorFlow Framework, a series of system testing was conducted using various variations of longan fruit images, totaling 360 samples. The aim of this research was to classify longan fruit types based on the extraction of color features from fruit images. The test results showed the highest accuracy rate reached 98.7% and an average accuracy of 89.6% on the train and test data with an 80%:20% ratio. The developed application successfully distinguished five categories of longan fruit, namely diamond river longan, itoh longan, mata lada longan, red longan, and pingpong longan. This study used a multi-class dataset as the data source. By using the KNN method with a parameter k=5, the system was able to classify longan fruit images with 78% accuracy in the 80%:20% train-validation data split scenario. These findings provide a positive perspective on the potential application of the KNN method in classifying longan fruit types based on the extraction of color features from fruit images. This research makes a significant contribution to the development of automatic recognition and classification systems for longan fruit using image processing techniques