An Open Source Framework for Real-Time, Incremental, Static and Dynamic Hand Gesture Learning and Recognition

Real-time, static and dynamic hand gesture learning and recognition makes it possible to have computers recognize hand gestures naturally. This creates endless possibilities in the way humans can interact with computers, allowing a human hand to be a peripheral by itself. The software framework developed provides a lightweight, robust, and practical application programming interface that helps further research in the area of humancomputer interaction. Approaches that have proven in analogous areas such as speech and handwriting recognition were applied to static and dynamic hand gestures. A semi-supervised Fuzzy ARTMAP neural network was used for incremental online learning and recognition of static gestures; and, Hidden Markov models for online recognition of dynamic gestures. A simple anticipatory method was implemented for determining when to update key frames allowing the framework to work with dynamic backgrounds.

• Type: Journal
• Year: 2009
• Where: HCI
• Authors: Todd C. Alexander, Hassan S. Ahmed, Georgios C. Anagnostopoulos
• Keywords: Dynamic Hand Gestures, Gestures, Hand Gestures, HCI 2009, Human Computer Interaction