Subsystem for Simple Dynamic Gesture Recognition Using 3DCNNLSTM

Abstract

The article describes the subsystem of the intellectual information-communication system developed at MTUCI. The analysis of the latest developments and research in the field of sign language recognition systems has been carried out. The net architecture and the training method have been considered for 3DCNNLSTM model that will be used to recognize simple dynamic gestures. A training data set, initial conditions, experiments performed and the use of learning outcomes are described.

Keywords

• 3DCNN
• LSTM
• FC
• 3DCNNLSTM
• simple dynamic gestures recognition