¹Department of Art, West Anhui University, Luan 237012, China.
²Department of Electronic Engineering, Fudan University, Shanghai, 200433, China.

---

Abstract—The hand gesture recognition is an effective way to apply human computer interaction which should aim at good recognition accuracy and high speed. To solve the problem of hand gesture recognition in complex scenes, we propose a new approach for hand gesture recognition which is based on the faster regional convolutional neural network (Faster-RCNN) deep learning algorithm with five layers of neural network. A benchmark database with gestures is used, several general characters under the complex environmental background are chosen to as the processing object. The model is established with a certain number samples for training. The results of model checking show that the algorithm can identify the gesture recognition categories effectively, quickly and accurately with low computational cost. The accuracy can reach highly up to 99.2% which is great significance for human computer interaction application.

Index Terms—Human computer interaction, hand gesture recognition, image processing, deep learning, faster regional convolutional neural network.

[PDF]

Cite: Xiaoguang Yu, Yafei Yuan, "Hand Gesture Recognition Based on Faster-RCNN Deep Learning," Journal of Computers vol. 14, no. 2, pp. 101-110, 2019.