Віртуальна клавіатура розпізнавання жестів
Loading...
Date
2020
Authors
Куценко, Михайло
Ющенко, Юрій
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Зроблено порівняння переваг і недоліків різних типів клавіатур і запропоновано принципово новий вид клавіатури: віртуальну клавіатуру розпізнавання жестів, що здатна налаштовуватись на вподобання користувачів, які за власним бажанням можуть навчити клавіатуру розпізнавати свої жести як натискання на ту чи іншу кнопку клавіатури.
У роботі описано розроблений прототип цієї віртуальної клавіатури й наведено висновки щодо зручності використання таких клавіатур.
Запропоновану технологію розпізнавання жестів і налаштування клавіатури може бути застосовано для створення інших пристроїв введення інформації жестами, наприклад, комп’ютерної миші, дігітайзера, джойстика чи будь-якого іншого ігрового контролера.
In this study we’ve considered advantages and disadvantages of different keyboard types and propose a fundamentally new one: keyboard based on gesture recognition. Keyboard itself is glove-shaped, and is supposed to be worn on hand. It provides enhanced mobility, as this keyboard doesn’t need any surface in order to type. Its compactness is a notable feature, since standard keyboards take up more space, which can cause an issue during a trip. Its technology is based on detecting flexion of the fingers and recognizing incoming readings as a press of a specific symbol. Powered with machine learning, it can adapt to users’ hand’s physiological features as well as to their way of using said keyboard. Its functionality can be extended by defining new symbols or making a set of commands to execute on triggering, and since metrics are range-based, they can be easily inserted in an existing set. Due to technology’s analog nature, many other input devices can be emulated using the base glove, with only the need for writing supporting drivers. A working prototype of such keyboard has been developed and its build and realization are described in this study. Prototype has been built on top of an Arduino UNO r3 board, using one of the most common types of LEDs and photoresistors, which are often shipped in starter kits. As well, a simple driver has been developed, which can correctly recognize input symbols and implement adaptation to specificities of user’s typing, using optimized k-nearest neighbors machine learning algorithm. Simplicity of its build, and availability of its components suggest low production cost and greater opportunity of serial production. As well, conclusions have been made about the convenience of its usage, and whether it meets expected requirements.
In this study we’ve considered advantages and disadvantages of different keyboard types and propose a fundamentally new one: keyboard based on gesture recognition. Keyboard itself is glove-shaped, and is supposed to be worn on hand. It provides enhanced mobility, as this keyboard doesn’t need any surface in order to type. Its compactness is a notable feature, since standard keyboards take up more space, which can cause an issue during a trip. Its technology is based on detecting flexion of the fingers and recognizing incoming readings as a press of a specific symbol. Powered with machine learning, it can adapt to users’ hand’s physiological features as well as to their way of using said keyboard. Its functionality can be extended by defining new symbols or making a set of commands to execute on triggering, and since metrics are range-based, they can be easily inserted in an existing set. Due to technology’s analog nature, many other input devices can be emulated using the base glove, with only the need for writing supporting drivers. A working prototype of such keyboard has been developed and its build and realization are described in this study. Prototype has been built on top of an Arduino UNO r3 board, using one of the most common types of LEDs and photoresistors, which are often shipped in starter kits. As well, a simple driver has been developed, which can correctly recognize input symbols and implement adaptation to specificities of user’s typing, using optimized k-nearest neighbors machine learning algorithm. Simplicity of its build, and availability of its components suggest low production cost and greater opportunity of serial production. As well, conclusions have been made about the convenience of its usage, and whether it meets expected requirements.
Description
Keywords
інтерфейс, клавіатура, миша, інтернет речей, IoT, розпізнавання, жести, стаття, interface, keyboard, mouse, Internet of things, IoT, artificial intelligence, machine learning, recognition, gestures
Citation
Куценко М. В. Віртуальна клавіатура розпізнавання жестів / Куценко М. В., Ющенко Ю. О. // Наукові записки НаУКМА. Комп'ютерні науки. - 2020. - Т. 3. - С. 138-140.