ZENG Qingxi,PENG Hui.Human Behavior Recognition based on ResNeXt-GRU and Cluster Sampling[J].Journal of Chengdu University of Information Technology,2022,37(01):40-45.[doi:10.16836/j.cnki.jcuit.2022.01.007]
基于ResNeXt-GRU和聚类采样的人体行为识别
- Title:
- Human Behavior Recognition based on ResNeXt-GRU and Cluster Sampling
- 文章编号:
- 2096-1618(2022)01-0040-06
- 关键词:
- 行为识别; 聚类; ResNeXt; 门控循环单元(GRU); Softmax
- Keywords:
- action recognition; clustering; ResNeXt; gate recurrent unit(GRU); Softmax
- 分类号:
- TP391
- 文献标志码:
- A
- 摘要:
- 为有效捕捉行为中的时序关系，增强网络的特征表达能力，提出一种基于ResNeXt-GRU的人体行为识别方法。首先，使用聚类算法提取行为视频关键帧序列，输入ResNeXt网络中进行空间维度上的特征提取。然后，将输出的特征向量全部输入门控循环单元(GRU)网络中进行时序学习。最后，利用Softmax分类器进行分类。在UCF101和HMDB51数据集上分别进行实验，识别准确率为93.7%和69.2%。实验结果表明与现有的其他许多行为识别方法相比，识别准确率得到了一定的提升。
- Abstract:
- In order to effectively capture the temporal relationships in behaviors and enhance the feature representation ability of the network, a human behavior recognition method based on ResNeXt-GRU is proposed. First of all, the behavioral video key frame sequences are extracted by using a clustering algorithm and then input to the ResNeXt network for feature extraction in spatial dimension. Then, the output feature vectors are all input into the gate recurrent unit(GRU)network for temporal learning. Finally, a Softmax classifier is used for classification. Experiments on UCF101 and HMDB51 datasets recognition accuracy of 93.7% and 69.2%, respectively. The experimental results show that the recognition accuracy was improved compared with many other existing behavior recognition methods.
参考文献/References:
[1] 凌佩佩,邱崧,蔡茗名,等.结合特权信息的人体动作识别[J].中国图象图形学报,2017,22(4):482-491.
[2] Simonyan K,Zisserman A.Two-stream convolutional networks for action recognition in videos[J].arXiv preprint arXiv:2014,1406:2199.
[3] Wang L,Xiong Y,Wang Z,et al.Temporal segment networks:Towards good practices for deep action recognition[C].European conference on computer vision.Springer,Cham,2016:20-36.
[4] Tran D,Bourdev L,Fergus R,et al.Learning spatiotemporal features with 3d convolutional networks[C].Proceedings of the IEEE international conference on computer vision.2015:4489-4497.
[5] Donahue J,Anne Hendricks L,Guadarrama S,et al.Long-term recurrent convolutional networks for visual recognition and description[C].Proceedings of the IEEE conference on computer vision and pattern recognition.2015:2625-2634.
[6] Zhao H,Jin X.Human action recognition based on improved fusion attention cnn and rnn[C].2020 5th International Conference on Computational Intelligence and Applications(ICCIA).IEEE,2020:108-112.
[7] WU X iru,XUE Ganggang.基于图像聚类的交通标志CNN快速识别算法[J].智能系统学报,2019,14(4):670-678.
[8] Hu H,Yang Y.A Combined GLQP and DBN-DRF for Face Recognition in Unconstrained Environments[C].2017 2nd International Conference on Control,Automation and Artificial Intelligence(CAAI 2017).Atlantis Press,2017:553-557.
[9] Xie S,Girshick R,Dollár P,et al.Aggregated residual transformations for deep neural networks[C].Proceedings of the IEEE conference on computer vision and pattern recognition.2017:1492-1500.
[10] He K,Zhang X,Ren S,et al.Deep residual learning for image recognition[C].Proceedings of the IEEE conference on computer vision and pattern recognition.2016:770-778.
[11] Cho K,Van Merriënboer B,Gulcehre C,et al.Learning phrase representations using RNN encoder-decoder for statistical machine translation[J].arXiv preprint arXiv:2014,1406:1078.
[12] Wang H,Schmid C.Action recognition with improved trajectories[C].Proceedings of the IEEE international conference on computer vision.2013:3551-3558.
[13] Qiu Z,Yao T,Mei T.Learning spatio-temporal representation with pseudo-3d residual networks[C].proceedings of the IEEE International Conference on Computer Vision.2017:5533-5541.
[14] Tran D,Ray J,Shou Z,et al.Convnet architecture search for spatiotemporal feature learning[J].arXiv preprint arXiv:2017,5038:1708.
[15] 蒋圣南,陈恩庆,郑铭耀,等.基于ResNeXt的人体动作识别[J].图学学报,2020,041(002):277-282.
[16] 陈颖,来兴雪,周志全,等.基于3D双流卷积神经网络和GRU网络的人体行为识别[J].计算机应用与软件,2020,37(5):164-168,218.
相似文献/References:
[1]鲍杨婉莹,蒋 瑜,李 冬.基于2型模糊集的粗糙模糊C-means算法[J].成都信息工程大学学报,2020,35(04):406.[doi:10.16836/j.cnki.jcuit.2020.04.007]
BAO Yangwanying,JIANG Yu,LI Dong.Type-2 Fuzzy Set based Rough Fuzzy C-means Clustering Algorithm[J].Journal of Chengdu University of Information Technology,2020,35(01):406.[doi:10.16836/j.cnki.jcuit.2020.04.007]
备注/Memo
收稿日期:2021-08-29
基金项目:四川省科技计划资助项目(2019YJ0356)