@article{article_1706643, title={VR-Assisted Comparative Analysis of HEVC Intra Prediction Modes Using Deep Learning and Rule-Based Systems}, journal={Black Sea Journal of Engineering and Science}, volume={8}, pages={1957–1966}, year={2025}, DOI={10.34248/bsengineering.1706643}, author={Kaplan, Mücahit}, keywords={HEVC, Intra prediction, Convolutional neural network, Rule-based expert system, VR visulation, Video coding}, abstract={This paper proposes an artificial intelligence-based solution to classify the three basic intrinsic prediction modes (Planar, DC and Angular) defined in the High Efficiency Video Coding (HEVC) standard. A convolutional neural network (CNN) based deep learning model trained with 32×32 blocks obtained from 30+ classical gray level test images is developed. As a result of the training, the model demonstrated a successful classification performance with an overall accuracy of over 89% and a macro F1 score of approximately 88%. The model was converted into ONNX format and integrated into a Unity-based virtual reality (VR) environment, thus creating an interactive analysis platform where users can observe the predictions of both artificial intelligence and rule-based systems at the block level comparatively. In this environment, users can also examine the reasoning of the predictions. The proposed system provides a holistic solution in terms of classification performance, interpretability and user experience, and makes innovative contributions to the analysis and visualization of video coding processes for educational purposes.}, number={6}, publisher={Karyay Karadeniz Yayımcılık Ve Organizasyon Ticaret Limited Şirketi}