GestARLite: An On-Device Pointing Finger Based Gestural Interface for Smartphones and Video See-Through Head-Mounts

TL;DR

This paper introduces a lightweight, real-time hand gesture recognition framework for wearable devices that combines deep learning models for localization and classification, enabling intuitive interaction in mixed reality applications.

Contribution

It presents a novel on-device gesture recognition system using a cascade of lightweight deep learning models optimized for smartphones and wearable MR devices.

Findings

Achieves 80% classification accuracy on EgoGestAR dataset

Operates in real-time with 0.12 seconds latency

Works effectively on frugal wearable devices like Google Cardboard

Abstract

Hand gestures form an intuitive means of interaction in Mixed Reality (MR) applications. However, accurate gesture recognition can be achieved only through state-of-the-art deep learning models or with the use of expensive sensors. Despite the robustness of these deep learning models, they are generally computationally expensive and obtaining real-time performance on-device is still a challenge. To this end, we propose a novel lightweight hand gesture recognition framework that works in First Person View for wearable devices. The models are trained on a GPU machine and ported on an Android smartphone for its use with frugal wearable devices such as the Google Cardboard and VR Box. The proposed hand gesture recognition framework is driven by a cascade of state-of-the-art deep learning models: MobileNetV2 for hand localisation, our custom fingertip regression architecture followed by a Bi-LSTM model for gesture classification. We extensively evaluate the framework on our EgoGestAR dataset. The overall framework works in real-time on mobile devices and achieves a classification accuracy of 80% on EgoGestAR video dataset with an average latency of only 0.12 s.