Momordica Charantia Size Classification Using Mask Region-Based Convolutional Neural Network

Bitter melon (Momordica charantia), also known locally as Ampalaya, holds significant economic and medicinal value, but its size classification remains a challenge due to its irregular morphology. Proper size classification is essential for maintaining packaging standards and ensuring efficient market distribution. Thus, this study presents a deep learning-based approach for automated bitter melon size classification using the Mask R-CNN model for instance segmentation, the Harris Corner Detection algorithm for contour detection, and the Douglas-Peuker algorithm for contour simplification. A hardware prototype was developed using a Raspberry Pi 4 Model B and a camera module to capture images for real-time processing. The system achieved an overall accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 7. 9 1 \%}$</tex> in classifying the sizes of bitter melons. The model displayed a high classification accuracy, demonstrating its potential for improving agricultural automation and postharvest processes.