Recent Advances in Vision-Based Hand Gesture Recognition for AI Virtual Mouse Systems: A Review

Hand gesture recognition has emerged as a promising approach for enabling intuitive and touchless human– computer interaction (HCI). Conventional input devices such as the mouse and keyboard limit natural interaction, particularly in environments where contactless operation is preferred. This paper presents the design and performance evaluation of a real-time AIbased virtual mouse system using vision-based hand gesture recognition. The proposed framework utilizes a lightweight Convolutional Neural Network (CNN) to classify hand gestures captured through a standard RGB webcam. A custom dataset consisting of approximately 6000 gesture images was collected under varying lighting conditions, backgrounds, and hand orientations to improve model generalization. The system supports four primary gestures corresponding to cursor movement, scrolling, left click, and right click operations. To enhance usability in real-time environments, cursor smoothing and click debouncing mechanisms are incorporated to reduce jitter and prevent unintended actions. Experimental evaluation demonstrates that the proposed CNN model achieves a classification accuracy of 97.9%, outperforming traditional machine learning algorithms including Support Vector Machine, K-Nearest Neighbors, Decision Tree, and Random Forest. The system operates at approximately 28–30 frames per second on standard CPU hardware, demonstrating its practicality for real-time human–computer interaction applications. The results indicate that the proposed framework provides a reliable and cost-effective solution for gesture-based touchless interfaces in assistive technology, public systems, and immersive computing environments.