Automated Conveyor System of Sorting and Grading for Red Chili Pepper (Capsicum annum L.) using Image Processing and Artificial Neural Network

This research aims to design an automatic sorting and grading tool driven by color sensor processed through image processing and artificial neural networks (ANN). The research stage consists of data collection in a Mini Studio, image processing using ImageJ, and image classification with ANN. The automatic sorting process begins with items entering the belt, where they are processed in four phases: (1) separating good and rejects chili, (2) separating red from green chili, (3) distinguishing large and small red peppers, and (4) separating large and small green peppers. Automatic sorting and grading were based on image data processed using ANN. The best activation function was tansig-logsig-purelin with MAPE 1.220, RMSE 0.010, and R² = 1 during training. During testing, the MAPE 0.158, RMSE 1.790, and R² = 0.963. The criteria produced grade 1 (red, 10-15 cm), grade 2 (green, 10-15 cm), grade 3 (red, 5-9.99 cm), and reject grade. The quality of large red chilies is used as a reference for market pricing: grade 1 (IDR 60,000/kg), grade 2 (IDR 40,000/kg), and grade 3 (IDR 25.000 – 35,000). Assessing quality based on color with an automatic conveyor can reduce sorting and grading time by 70% compared to conventional methods. Keywords: ANN, Color, Grading, Image Processing, Sorting.