Real Time Hand Gesture Recognition In Industry

novembre 2021
Ingénierie & Outils numériques
Communications avec actes dans un congrès international
Auteurs : William Dumoulin (Sans affiliation), Nicolas Thiry (Sans affiliation), Rim Slama (FORS)
Conférence : International Conference on Video, Signal and Image Processing, 17 novembre 2021

With the 4th industrial revolution and the increaseduse of cobots in the industries comes many opportunities for newgeneration control panels. In this article, we proposed to developa deep learning model to recognise in real time 10 differentgestures that can be used to interact with a cobot. We proposeda new dataset containing gestures that can be used in industrialcontext. The videos were taken from a computer webcam andthen processed to remove the noise created by the background byisolating the movement of the gray scale images. We proposedto extract the spatio-temporal features by the combination of3D convolution and LSTM layers. We also proposed a real timemethod to recognize our gestures. Our experimental results showfor 8 out of 10 gestures, a recognition rate of more than 90%.Furthermore, an interface was created to test our method in realtime and to add a new class of gestures to be recognized by ourmodel.