Real-time Dynamic Hand Gesture Learning and Classification using Fuzzy ARTMAP

The all materials are provided in Korean

Abstract

Previously proposed models for time series data such as RNN and LSTM have limitations in learning new categories in real time. This paper uses Fuzzy ARTMAP, which enables a model to quickly learn new categories with a small amount of computation time. In addition, This paper proposes a method for converting static hand joint data into dynamic information for learning and classifying dynamic hand gestures

Paper (ICROS 2020, Best undergraduate paper)

Presentation

Method

Results(Hand shape)

You can find the test video(Hand shape) from Here

Results(Hand Position)

You can find the test video(Hand Position) from Here

Fuzzy ARTAMP Tuturial

You can learn the process of Fuzzy ARTAMP algorithm from Fuzzy ARTAMP Tuturial

You also can watch the videos for this algorithm (Korean)

Comment

Fuzzy ARTMAP을 이용한 Dynamic Hand Gesture 실시간 학습 및 인식

-class_ARTAMP.py

-class_GUI_dynamic.py

-class_server.py

-main.py

(카메라 센서 코드는 업로드 하지 않음)

실시간 손 동작 학습 및 인식을 위한 코드

교수님 지도에 항상 감사하고 있습니다.