Air Writing Detection and Recognition

Abstract

Air Writing, a groundbreaking concept, revolutionizes the act of writing by allowing users to create characters or words in free space through hand or finger movements and coloured light. Unlike traditional pen-and-paper methods, this approach replaces pen-up and pen-down motions with colour shifts or light toggling to indicate the beginning and end of characters or words. The Air Handwriting Recognition project combines computer vision object tracking with handwriting recognition using machine learning techniques. Using a computer's webcam, the system tracks the characters users write in the air, following a user-selected colour with the aid of a mask. These tracked movements are then transcribed onto a virtual canvas, mimicking a plain whiteboard. The resulting canvas image serves as input for the recognition model, employing machine learning to interpret air-written words and characters. The integration of colour-based tracking and advanced recognition algorithms ensures the avoidance of plagiarism, making Air Handwriting Recognition a cutting-edge solution for hands-free writing in the digital realm.. A brief history of CNN and other approaches to characters detection and recognition are discussed in this paper