<![CDATA[The development of generative Artificial Intelligence (AI) technology enables computer systems to produce highly realistic images that closely resemble real photographs. This condition creates challenges in distinguishing AI-generated images from real images visually. This research aims to develop an image classification system capable of distinguishing AI-generated images and real images using visual feature extraction and ensemble learning algorithms.The research method consists of several stages including image preprocessing by resizing images to 256 × 256 pixels, visual feature extraction including RGB color histogram, grayscale intensity distribution, texture features using Gray Level Co-occurrence Matrix (GLCM), and edge features using the Canny Edge Detection method. The extracted features are then used as input for several classification algorithms such as Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest. Furthermore, model combination is performed using an ensemble learning method with a hard voting technique.The experimental results show that the Random Forest model achieved an accuracy of 65.71%, while the ensemble learning method achieved an accuracy of 65.00% with an F1-score of 0.6918. The developed system is also implemented as a web-based application using the Streamlit framework, allowing users to upload images and obtain prediction results directly. The results indicate that the combination of visual feature extraction and ensemble learning can be used as an approach to help identify AI-generated images and real images.]]>
Copyrights © 2026
