Human-Computer Interaction (HCI) has evolved from traditional input devices to intuitive, contactless interfaces. This paper presents an AI-driven hand gesture recognition system enabling touchless computer control via a standard webcam. The system supports operations such as cursor movement, clicking, scrolling, volume and brightness control, screenshot capture, screen recording, and system shutdown through predefined hand gestures. A novel automated zoom feature, powered by you-only-look-once v8 (YOLOv8), enhances gesture detection accuracy by adjusting the camera focus according to the user’s distance. Leveraging MediaPipe for hand tracking, the system achieves real-time performance on standard hardware. The automated zoom mechanism dynamically adapts to user proximity, raising the average recognition accuracy for five representative gestures at 2 m from 60.91% to 93.01%, a significant leap for accessibility, remote work, and hygienic interfaces. Experimental results highlight substantial accuracy gains with smart adaptive zoom, especially for complex two-hand gestures.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

AI-Driven Hand Gesture Recognition with Adaptive Zoom for Smart Proximity-Based Computer Interaction

  • Dubacharla Gyaneshwar,
  • Sri Likhith Gorrela

摘要

Human-Computer Interaction (HCI) has evolved from traditional input devices to intuitive, contactless interfaces. This paper presents an AI-driven hand gesture recognition system enabling touchless computer control via a standard webcam. The system supports operations such as cursor movement, clicking, scrolling, volume and brightness control, screenshot capture, screen recording, and system shutdown through predefined hand gestures. A novel automated zoom feature, powered by you-only-look-once v8 (YOLOv8), enhances gesture detection accuracy by adjusting the camera focus according to the user’s distance. Leveraging MediaPipe for hand tracking, the system achieves real-time performance on standard hardware. The automated zoom mechanism dynamically adapts to user proximity, raising the average recognition accuracy for five representative gestures at 2 m from 60.91% to 93.01%, a significant leap for accessibility, remote work, and hygienic interfaces. Experimental results highlight substantial accuracy gains with smart adaptive zoom, especially for complex two-hand gestures.