AI Gesture Control: Touchless Navigation for Web Applications Using Deep Learning

Abstract

The use of AI-based gesture control in web systems has evolved as a futuristic technology to provide richer user interaction and accessibility by facilitating touchless interaction and navigation. Progress has been notable in the field in the last decade through improvements in deep learning techniques, particularly CNNs and RNNs. In initial studies, hand gesture recognition was aimed for simple interaction, but factors such as uncertainty of the environment, limitations of devices, and latency were important considerations. Hybrid models combined with 3D gesture recognition have drawn interest in further studies with greater accuracy and capability to accommodate richer gestures. Despite the use of such systems, however, there has been restraint in using these in real-time web applications due to factors concerning hardware dependence, real-time capability, and scalability. Despite these advances, a number of research gaps remain. Firstly, the application of AI-powered gesture recognition on mainstream web platforms is still not sufficiently researched, particularly with regards to cross-device compatibility. The majority of current systems rely on specific hardware like depth cameras, thereby restricting their utility for the masses. Secondly, although multi-modal voice and gesture interfaces have shown promise, their viability in practical usage, particularly in noisy environments, remains to be fine-tuned. Thirdly, interfacing such technology across different domains, for example, health, e-business, and learning, needs to factor in user diversity, privacy, and personalization needs for gesture sets. Filling these gaps could lead to the creation of more robust, inclusive, and scalable gesture-controlled web application systems that eventually benefit user experiences across different domains.