Paper Title
Optimizing Pomegranate Sorting Through Multi-Feature Extraction and ML Techniques

Abstract
The sorting of pomegranates is a crucial step in their collection and packing. Previously, this task was performed manually by laborers who sorted the fruit based on quality. However, manual sorting is extremely labor-intensive, expensive, and lacks the precision required for export markets. This method is also constrained by time and accuracy, which has driven the development of automated sorting systems that offer faster and more precise results. This research applies Contrast Limited Adaptive Histogram Equalization (CLAHE) and background removal techniques during the preprocessing stage to enhance image quality and remove irrelevant information. After preprocessing, several feature extractors, including ORB (Oriented FAST and Rotated BRIEF), SIFT (Scale-Invariant Feature Transform), LBP (Local Binary Patterns), HOG (Histogram of Oriented Gradients), and Gabor filters, are employed to generate datasets. These feature sets are then used to train various Machine Learning (ML) models, such as KNN, SVM, RFC, and ANN. Among the different feature extraction methods, LBP achieved the highest accuracy at 98.52%, followed by Gabor filters with 97.68% accuracy. Keywords - ANN, Feature Extraction, ORB, CLAHE, SIFT, SURF, LBP