Performance Analysis of Dimensionally Reduced Models in Seed Classification

Abstract

Accurate identification of seed varieties is essential for improving crop yield, maintaining genetic purity, and supporting precision agriculture practices. This study presents a data-driven approach for seed variety recognition using machine learning techniques combined with systematic feature reduction and comprehensive model performance analysis. A dataset comprising morphological, textural, and color-based attributes of seeds was preprocessed to remove noise and redundancy. Dimensionality reduction and feature selection methods such as Principal Component Analysis (PCA), Recursive Feature Elimination (RFE), and correlation-based filtering were employed to identify the most discriminative features.Multiple classification models, including Support Vector Machines (SVM), Random Forest (RF), k-Nearest Neighbors (KNN), and Logistic Regression (LR), were trained and evaluated using performance metrics such as accuracy, precision, recall, F1-score, and confusion matrix analysis. Comparative results demonstrate that optimized feature subsets significantly enhance classification performance while reducing computational complexity. Among the evaluated models, ensemble-based approaches achieved superior accuracy and robustness across different seed varieties.The proposed framework highlights the effectiveness of integrating feature reduction strategies with model benchmarking to develop a reliable and scalable seed variety recognition system. This approach can assist agricultural stakeholders in automated seed grading, quality control, and smart farming applications