|
ИСТИНА |
Войти в систему Регистрация |
ИСТИНА ПсковГУ |
||
EVALUATING ALGORITHMS FOR THE AUTOMATIC DETECTION OF HAND GESTURE PHASE BOUNDARIESтезисы доклада
- Авторы: Serbina A.P., Nikolaeva Y.V.
- Сборник: Когнитивная наука в Москве: новые исследования. Материалы конференции 25 – 26 июня 2025 г
- Тезисы
- Год издания: 2025
- Издательство: ООО "Буки-Веди"
- Местоположение издательства: Москва
- Аннотация: This research tests the automatic annotation of hand gestures. It presents a comparative analysis of different algorithms based on tracking formal signs of hand movements using the MediaPipe library. The aim of the study is to evaluate these algorithms in identifying gesture phase boundaries. The methodology involves analyzing a 22-second video fragment from the multimodal corpus “RUPEX”. For each frame of the video, we obtained the coordinates of the tips of the thumb and index finger of both hands. The algorithms were evaluated on the metrics of precision and recall. The results showed that tracking external motion features can identify gesture phase boundaries with an accuracy of 100 ms, but none of the algorithms demonstrated both high precision and high recall.
- Добавил в систему: Николаева Юлия Владимировна