Aprilia Melani
Universitas Dinamika Bangsa

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Classification of Spice Images Using a Combination of HOG and Color Histogram Features with a Machine Learning Method Aprilia Melani; M. Irwan Bustami; Desi Kisbianty
Jurnal PROCESSOR Vol 21 No 2 (2026): Jurnal Ilmiah Processor, Edisi Oktober 2026
Publisher : LPPM Universitas Dinamika Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33998/processor.2026.21.2.2800

Abstract

Classification of object images with similar visual characteristics is one of the challenges in digital image processing. Spices have variations in shape, texture, and color that are relatively similar between types, making the visual identification process difficult. Therefore, an image processing and machine learning-based approach is needed to assist the classification process more objectively and consistently. The aim of this study is to develop various machine learning algorithms to classify spice images and compare their performance by combining Histogram of Oriented Gradients (HOG) and Color Histogram features. The data source used is Mendeley Data, which consists of eight classes of spices with an initial total of 4,134 images. To address class distribution imbalance, data augmentation was performed, resulting in 8,800 images. The research stages include image pre-processing, feature extraction, dimensionality reduction using Principal Component Analysis (PCA), as well as the development of classification models using Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and XGBoost algorithms optimized with GridSearchCV. Model performance is evaluated based on accuracy, precision, recall, F1-score, and confusion matrix metrics. The test results showed that the SVM algorithm had the best performance with an accuracy rate of 88%, followed by XGBoost with 84% and KNN with 81%. These results indicate that the combination of HOG and Color Histogram features effectively represents the visual characteristics of spices, and the SVM algorithm is able to optimally handle high inter-class similarity.