Enhanced Bot Game for a Massively Multiplayer Online Role-Playing Game (MMORPG) Using Object Detection

The increasing prevalence of the internet has fueled the global expansion of massively multiplayer online role-playing games (MMORPGs), attracting a vast player base. However, players often face time constraints while striving to progress in these immersive virtual worlds. To address this challenge, various techniques have been explored, including Template Matching, Color Segmentation, Canny Edge Detection, and deep learning models, to develop bot games aimed at assisting players. In this study, we evaluated the effectiveness of several object detection techniques, namely Template Matching, Color Segmentation, Canny Edge Detection, Cascade Classifier, and YOLOv5, within the context of MMORPGs, focusing on the Dragonica game. Through experimentation and analysis, we found that YOLOv5 outperformed other techniques in terms of object detection accuracy and efficiency. While Template Matching exhibited proficiency in detecting stationary objects with consistent attributes, it proved inadequate for dynamic environments characterized by moving objects. Our findings underscore the importance of comprehensive model training across diverse game scenarios and the incorporation of instance object detection to enhance precision. Additionally, we highlight the need for further research to optimize object detection methodologies in MMORPGs, aiming for robust and generalized solutions to cater to the evolving dynamics of virtual worlds. Overall, this study provides insights into the efficacy of different object detection techniques in MMORPGs, offering a roadmap for future advancements in this domain.