AI VIRTUAL PRESENTATION SYSTEM: REAL-TIME TOUCHLESS SLIDE NAVIGATION USING DEEP LEARNING

Traditional presentation systems rely on physical devices such as keyboards, mice, or remote clickers,
which restrict presenter mobility and reduce interaction quality. This paper pro- poses an AI Virtual
Presentation System that enables real-time touchless slide navigation using Convolutional Neural
Networks (CNNs). The system performs gesture recognition using infrared- based hand image datasets
and integrates the trained model into a full-stack web application. Unlike browser-based AI solutions,
the proposed architecture shifts computation to a FastAPI backend using WebSockets for ultra-low
latency communication. The system ensures efficient gesture classification, reduced client- side load,
and seamless presentation control through webcam input. Experimental results demonstrate high
accuracy, improved user experience, and real-time responsiveness, making the system suitable for
professional and academic environments.