REAL-TIME HAND GESTURE DETECTION AND INTERPRETATION ON ANDROID
International Research Journal of Modernization in Engineering Technology and Science · 2024
This paper focuses on developing a camera application for Android devices that leverages TensorFlow Lite to continuously classify hand gestures in real-time.Utilizing the device's front camera, the app captures frames and identifies predefined gestures through a trained TensorFlow Lite model.The app is designed to be userfriendly and efficient, requiring minimal setup from the user's end.The TensorFlow Lite model, essential for gesture recognition, is seamlessly integrated into the application through Gradle scripts during the build process, eliminating the need for manual downloads or setup.The model's training is based on a diverse set of hand gesture images to ensure high accuracy and robustness in classification.For optimal performance and compatibility, the application is tailored for Android devices meeting specific requirements, including a minimum operating system of Android 6.0 (Marshmallow) and enabled developer mode.The development and testing of the app have been conducted using Android Studio IDE, with version 2021.2.1 (Chipmunk) being the recommended environment.Building the app involves a straightforward process in Android Studio, where developers are guided through opening the project, synchronizing Gradle, and deploying the app to a connected Android device in developer mode.Additionally, the app offers customization options for the TensorFlow Lite model, allowing developers to enhance and tailor the gesture classification capabilities to specific needs.This application stands as a practical demonstration of integrating TensorFlow Lite into Android development, showcasing the potential of machine learning in enhancing user interaction and experience through gesture recognition.