Перспективи діагностики рухових розладів за допомогою методів комп’ютерного зору на базі мобільного пристрою

The research work investigated markerless video analysis technologies based on the analysis of the relative position of heterogeneous parts of object images in successive frames. Modern video motion capture systems are ready-made clusters of points in the form of plates with four markers for long li...

Повний опис

Збережено в:
Бібліографічні деталі
Дата:2024
Автори: Андрущенко, М.А., Селіванова , К.Г.
Формат: Стаття
Мова:Ukrainian
Опубліковано: Vinnytsia National Technical University 2024
Теми:
Онлайн доступ:https://oeipt.vntu.edu.ua/index.php/oeipt/article/view/734
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Optoelectronic Information-Power Technologies

Репозитарії

Optoelectronic Information-Power Technologies
Опис
Резюме:The research work investigated markerless video analysis technologies based on the analysis of the relative position of heterogeneous parts of object images in successive frames. Modern video motion capture systems are ready-made clusters of points in the form of plates with four markers for long limb segments, a ‘cap’ with markers for the head, etc. The main models used in mobile devices to estimate the movements of the upper limbs and the biomechanics of joint movement in real time were studied, namely PoseNet, MoveNet Thunder, MoveNet Lightning, and BlazePose in Light, Full, and Heavy versions. The models were evaluated for key characteristics such as speed, device impact, and support for hardware acceleration.