ANALYSIS OF QUALITY CLASSIFICATION RESULTS OF CAYALE CHILI USING SUPPORT VECTOR MACHINE METHOD

Abstract

This research aims to implement the Support Vector Machine (SVM) method in classifying cayenne pepper images based on their quality. A total of 800 images of cayenne pepper were collected and grouped into four quality classes, namely rotten, greenish, dry and ripe. The classification process using SVM involves a preprocessing stage, image segmentation with Canny edge detection, and model performance evaluation. Implementation is carried out through the development of a web application with several worksheets, including model training worksheets, classification process, classification results, evaluation and export. System testing involves alpha and beta functional testing. Alpha functional testing includes homepage display and navigation tests, model training process, image classification process, classification results, classification evaluation, and export of classification results. Beta functional testing is carried out by involving users who provide feedback through questionnaires. The test results show that the application succeeded in classifying the image of cayenne pepper with high accuracy and received a positive assessment from users. The results of the F1-Score calculation for SVM classification show good model performance, with F1-Score values ​​for each quality class of cayenne pepper as follows: Rotten (0.973), Greenish (0.979), Dry (1.0), and Ripe (0.972). Thus, this research contributes to the application of SVM for image classification of cayenne peppers with satisfactory results, as well as presenting a reliable application in supporting the quality analysis of cayenne peppers.