Enhancing augmented reality experiences through advanced computer vision techniques

Abstrak

This study investigates the integration of advanced computer vision techniques, specifically semantic segmentation, depth estimation, and object detection, into augmented reality (AR) systems to enhance user immersion, contextual interaction, and real-world adaptability. The research aims to address the limitations of existing AR systems, which often struggle with real-time performance and context-aware interaction in dynamic environments. Using a design-based research (DBR) approach, the study involved testing the proposed AR system on mobile devices under controlled and real-world conditions. Performance was assessed using specific metrics: frame rate (FPS), inference latency, semantic accuracy, and user experience. The sample included 20 participants, recruited through convenience sampling, with inclusion criteria focused on individuals familiar with mobile AR applications. Ethical approval was obtained for the study, and informed consent was provided by all participants. The testing involved both simulation trials and real-world prototyping in varied environmental conditions, such as indoor and outdoor settings with different lighting and motion dynamics. Results indicate that the integrated of the vision stack significantly enhanced scene understanding and enabling stable, context-aware digital overlays, with the system maintaining a real-time frame rate of >27 FPS. User feedback, measured through a 5-point Likert scale survey, confirmed improved immersion, visual coherence, and satisfaction compared to baseline AR systems, with an average increase of 35% in perceived realism. The analysis also revealed that the system's performance remained consistent across varying environmental conditions, with minimal latency (less than 300ms) in dynamic re-anchoring of AR elements. Statistical tests (paired t-tests) confirmed the significance of these improvements, with p-values < 0.05 for all key metrics. This research contributes a scalable framework that bridges artificial intelligence, user experience design, and mobile AR deployment. It provides empirical evidence supporting the integration of computer vision techniques into AR systems, with practical implications for applications in education, healthcare, and industry. Future work will focus on expanding the user base, exploring hardware compatibility, and investigating multimodal AR interactions.