Recognition of Hand Gestures Using CNN

In this work, we propose a convolution neural network (CNN) method to recognize hand gestures of human task activities from a sequence of images/frames. The gesture is body language that humans use to express emotion and thoughts. The varied gestures of the five fingers and palm may have their physical meanings. Hand gesture recognition is a complicated system that is composed of gesture modeling, gesture analysis and recognition, and machine learning. This work is a CNN-based human hand gesture recognition system. CNN is a research branch of neural networks. Using a CNN to learn human gestures, there is no need to develop complicated algorithms to extract image features and learn them. Through the convolution and sub-sampling layers of a CNN, invariant features are extracted automatically. In our experiment, we provided a validation of the proposed method on recognizing human gestures which show robust results with various hand positions and orientations. Our experimental evaluation of four subjects performing four-hand gestures is implemented using Sebastien Marcel Dynamic Hand Posture Database.

Keywords: Human gesture recognition, Convolution Neural Network (CNN).