Research Of Gesture Recognition Based On Densely Connected 3DCNN And Convolutional GRU

Gesture recognition based on visual information is an active research topic.Gesture recognition technology has been applied in many fields,including virtual reality,augmented reality,sign language recognition,robot control,intelligent household,games and entertainment,medical auxiliary system,etc..With the advancement of intelligent,it has a huge potential application prospect.Gesture is a short form of physical activity that includes apparent information and movement information.Complex background,different lighting conditions,and the difference between individual gestures can affect the recognition of gestures.The gesture recognition based on video may be more difficult to recognize because of the poor quality of video and the relatively small of gestures.In this paper,the gesture recognition based on SKIG RGB-D multimodal isolated gesture video is studied.The RGB and Depth videos are extracted into the form of images.Then the sampled 32 frames from images are input to the densely connected 3DCNN component to learn short-term spatiotemporal features,after that the features input to the convolution GRU to learn long-term spatiotemporal features,finally the trained networks for single modal are used to multimodal fusion to improve the recognition accuracy.In this paper,99.07%recognition accuracy is obtained on the SKIG dataset,which achieves high accuracy and proves the validity of the network model proposed in this paper.