三维循环密集卷积神经网络在视频手势识别的应用
|
摘要
手势识别是当前计算机视觉的一个重要研究课题,由于手势旋转,角度等因素的影响,视频手势识别仍是一项具有挑战性的任务。该文提出了一种基于三维密集卷积神经网络和门限循环单元的双通道手势识别算法,通过三维密集卷积神经网络获取手势的空间信息,使用门限循环单元学习视频中手势的时序信息,最后融合RGB图像和深度图像的深度学习模型特征以此对手势进行识别。在ISOGD数据集上的实验表明,该手势识别算法能够有效提高了视频手势识别的准确率。
Recent vedio gesture recognition is an important research topic in computer vision,which is an still a challenging task due to the influence of gesture rotation, angle and other factors. In this paper, a two-channel gesture recognition algorithm based on 3-D dense convolution neural network and threshold cycle module is proposed. We acquire the spatial information of gesture by 3-D dense convolution neural network,get the temporal information of gesture in video by gated recurrent unit, and the deep learning model features of RGB image and depth image are fused to recognize gesture.The experiments on ISOGD datasets show that this gesture recognition algorithm can effectively improve the accuracy of video gesture recognition.
Recent vedio gesture recognition is an important research topic in computer vision,which is an still a challenging task due to the influence of gesture rotation, angle and other factors. In this paper, a two-channel gesture recognition algorithm based on 3-D dense convolution neural network and threshold cycle module is proposed. We acquire the spatial information of gesture by 3-D dense convolution neural network,get the temporal information of gesture in video by gated recurrent unit, and the deep learning model features of RGB image and depth image are fused to recognize gesture.The experiments on ISOGD datasets show that this gesture recognition algorithm can effectively improve the accuracy of video gesture recognition.
引文
[1]PIYUSH K,SIDDHARTH S R,and Anupam A.Hand data glove:A new generation real-time mouse for humancomputer interaction[C].International Conference on Recent Advances in Information Technology,Dhanbad,Jharkand,India,2012:750-755
[2]Wan J,Li S Z,Zhao Y.ChaLearn Looking at People RGB-DIsolated and Continuous Datasets for Gesture Recognition[C].2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops.IEEE,2016.
[3]Wang P,Li W,S.Liu,Z.Gao,et al.Large-scale isolated gesture recognition using convolutional neural networks[C].In IEEE International Conference on Pattern Recognition Workshops,2016.
[4]马杰,张绣丹,杨楠.融合密集卷积与空间转换网络的手势识别方法[J].电子与信息学报,2018,40(4):951-956.
[5]Zhu G,Zhang L,Mei L,et al.Large-scale Isolated Gesture Recognition using pyramidal 3D convolutional networks[C].2016 23rd International Conference on Pattern Recognition(ICPR).IEEE,2016.
[6]Zhu G,Zhang L,Shen P,et al.Multimodal Gesture Recognition Using 3-D Convolution and Convolutional LSTM[J].IEEE Access,2017,5:4517-4524.
[7]Huang G,Liu Z,Laurens V D M,et al.Densely Connected Convolutional Networks[J].2016.
[8]Tran D,Bourdev L,Fergus R,et al.Learning Spatiotemporal Features with 3D Convolutional Networks[J].2014.
[9]Kalchbrenner N,Danihelka I,Graves A.Grid Long Short-Term Memory[J].Computer Science,2015.
[2]Wan J,Li S Z,Zhao Y.ChaLearn Looking at People RGB-DIsolated and Continuous Datasets for Gesture Recognition[C].2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops.IEEE,2016.
[3]Wang P,Li W,S.Liu,Z.Gao,et al.Large-scale isolated gesture recognition using convolutional neural networks[C].In IEEE International Conference on Pattern Recognition Workshops,2016.
[4]马杰,张绣丹,杨楠.融合密集卷积与空间转换网络的手势识别方法[J].电子与信息学报,2018,40(4):951-956.
[5]Zhu G,Zhang L,Mei L,et al.Large-scale Isolated Gesture Recognition using pyramidal 3D convolutional networks[C].2016 23rd International Conference on Pattern Recognition(ICPR).IEEE,2016.
[6]Zhu G,Zhang L,Shen P,et al.Multimodal Gesture Recognition Using 3-D Convolution and Convolutional LSTM[J].IEEE Access,2017,5:4517-4524.
[7]Huang G,Liu Z,Laurens V D M,et al.Densely Connected Convolutional Networks[J].2016.
[8]Tran D,Bourdev L,Fergus R,et al.Learning Spatiotemporal Features with 3D Convolutional Networks[J].2014.
[9]Kalchbrenner N,Danihelka I,Graves A.Grid Long Short-Term Memory[J].Computer Science,2015.