Integrated Application Model for Hand Gesture Recognition

Abstract

Hand gesture-based human-computer interaction is both intuitive and versatile, with diverse applications such as in smart homes, games, operating theaters, and vehicle infotainment systems. An effective human-computer interaction system is required a good accuracy rate of recognition and speed. In our work, we have proposed a system model for static hand gesture recognition by using multiple common features. There are three contributions in this model:(1) A multiple features classification based on the Non-Dominated Sorting Genetic Algorithm II (NSGAII). The use of NSGAII can be reduced redundant features and minimizing feature value which effective on the execution cost of the system. (2) Proposed a new methodology of multiple features convolutional neural network (MFCNN) model to recognize both common and real-time hand gestures. (3) The generation of sequence sentences based on the Beam Search (BS) algorithm. Data of image labels that were received from the recognition stage combine with the CNN/Dailymail dataset is used to generate sentences.