Online Human Action Detection Using Joint Classification-Regression Recurrent Neural Networks | |
Li, Yanghao ; Lan, Cuiling ; Xing, Junliang ; Zeng, Wenjun ; Yuan, Chunfeng ; Liu, Jiaying | |
2016 | |
关键词 | Action detection Recurrent neural network Joint classification-regression SHORT-TERM-MEMORY |
英文摘要 | Human action recognition from well-segmented 3D skeleton data has been intensively studied and has been attracting an increasing attention. Online action detection goes one step further and is more challenging, which identifies the action type and localizes the action positions on the fly from the untrimmed stream data. In this paper, we study the problem of online action detection from streaming skeleton data. We propose a multi-task end-to-end Joint Classification-Regression Recurrent Neural Network to better explore the action type and temporal localization information. By employing a joint classification and regression optimization objective, this network is capable of automatically localizing the start and end points of actions more accurately. Specifically, by leveraging the merits of the deep Long Short-Term Memory (LSTM) subnetwork, the proposed model automatically captures the complex long-range temporal dynamics, which naturally avoids the typical sliding window design and thus ensures high computational efficiency. Furthermore, the sub-task of regression optimization provides the ability to forecast the action prior to its occurrence. To evaluate our proposed model, we build a large streaming video dataset with annotations. Experimental results on our dataset and the public G3D dataset both demonstrate very promising performance of our scheme.; CPCI-S(ISTP); lyttonhao@pku.edu.cn; culan@microsoft.com; jlxing@nlpr.ia.ac.cn; wezeng@microsoft.com; cfyuan@nlpr.ia.ac.cn; liujiaying@pku.edu.cn; 203-220; 9911 |
语种 | 英语 |
出处 | 14th European Conference on Computer Vision (ECCV) |
DOI标识 | 10.1007/978-3-319-46478-7_13 |
内容类型 | 其他 |
源URL | [http://ir.pku.edu.cn/handle/20.500.11897/460046] |
专题 | 信息科学技术学院 |
推荐引用方式 GB/T 7714 | Li, Yanghao,Lan, Cuiling,Xing, Junliang,et al. Online Human Action Detection Using Joint Classification-Regression Recurrent Neural Networks. 2016-01-01. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论