Green Chilli Quality Classification Using Deep Learning Models

This research uses deep learning models to propose an automatic system that classifies green chilies into three grades, namely Grade 1, Grade 2, and Grade 3, based on visual features like texture, color, size, and shape. It aims to help exporters achieve consistent quality, reduce waste, and meet international standards. Using a dataset of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2, 1 0 0}$</tex> images, several deep learning models, including MobileNet, DenseNet121, ResNet50, and VGG16, were evaluated. MobileNet showed the best results, achieving <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 6. 6 7 \%}$</tex> accuracy with a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{7 0: 3 0}$</tex> train-test split and a batch size of 16, due to its lightweight design and reliability. The study highlights the potential of such systems to simplify large-scale sorting, with future work focusing on real-time implementation and further improvements.