Object Classification Model Using Ensemble Learning with Gray-Level Co-Occurrence Matrix and Histogram Extraction

TL;DR

This paper presents an ensemble learning approach combining voting and combined classifiers to improve object classification accuracy using gray-level co-occurrence matrix and histogram features, achieving high performance metrics.

Contribution

It introduces a novel ensemble classification method that enhances object recognition accuracy in image processing by combining multiple classifiers and voting strategies.

Findings

Ensemble voting achieved 92.4% accuracy.

Combined classifier achieved 99.3% accuracy.

Methods significantly improve classification performance.

Abstract

In the field of object classification, identification based on object variations is a challenge in itself. Variations include shape, size, color, and texture, these can cause problems in recognizing and distinguishing objects accurately. The purpose of this research is to develop a classification method so that objects can be accurately identified. The proposed classification model uses Voting and Combined Classifier, with Random Forest, K-NN, Decision Tree, SVM, and Naive Bayes classification methods. The test results show that the voting method and Combined Classifier obtain quite good results with each of them, ensemble voting with an accuracy value of 92.4%, 78.6% precision, 95.2% recall, and 86.1% F1-score. While the combined classifier with an accuracy value of 99.3%, a precision of 97.6%, a recall of 100%, and a 98.8% F1-score. Based on the test results, it can be concluded that the use of the Combined Classifier and voting methods is proven to increase the accuracy value. The contribution of this research increases the effectiveness of the Ensemble Learning method, especially the voting ensemble method and the Combined Classifier in increasing the accuracy of object classification in image processing.