Gesture Recognition Based on Deep Learning in Complex Scenes
Nan Wang, Zhou Zhigang, Huan Lei, Jingqi Ma, Zhuang Jiajun, Duan Guangxue · 2019
Aiming at the problems of low recognition rate and robustness of the gesture recognition in complex scenes such as light changes and background interference, a gesture recognition algorithm based on Faster R-CNN is proposed, which is envoled from CNN. Applied the Faster R-CNN framework in objects recognition and classification algorithm in deep learning, to recognize and classify the gesture images collected in complex scenes, we adjust the key parameters of the Faster R-CNN framework to achieve the purpose of simultaneously detecting and recognizing gestures. Compared with the traditional gesture recognition method, the target detection and classification algorithm based on deep learning method in the complex scene has a significant improvement in the recognition accuracy, and the average recognition rate reaches 98.19%, the proposed algorithm avoids the over-fitting problem of the training model and has higher recognition accuracy and stronger robustness than the traditional algorithm.