Temporal action proposal generation aims to output the starting and ending times of each potential action for long videos and often suffers from high computation cost. To address the issue, we propose a new temporal convolution network called Multipath Temporal ConvNet (MTCN). In our work, one novel high performance ring parallel architecture based is further introduced into temporal action proposal generation in order to respond to the requirements of large memory occupation and a large number of videos. Remarkably, the total data transmission is reduced by adding a connection between multiplecomputing load in the newly developed architecture. Compared to the traditional Parameter Server architecture, our parallel architecture has higher efficiency on temporal action detection tasks with multiple GPUs. We conduct experiments on ActivityNet-1.3 and THUMOS14, where our method outperformsother state-of-art temporal action detection methods with high recall and high temporal precision. In addition, a time metric is further proposed here to evaluate the speed performancein the distributed training process. 相似文献
This paper investigates shortcomings that limit the performance of optical code division multiple access (OCDMA) systems including the low cardinality and data rate as well as the high power at reception. The main drawback for such systems known as multiple access interference accompanying by phase induced intensity noise is also investigated to effeciencly propose a novel two dimensional cyclic shift (2D-CS) code to be implemented in non-coherent OCDMA systems. The developed code is based on a one dimensional cyclic shift (1D-CS) code previously provided by research works processing spectral amplitude coding for optical code division multiple access (SAC-OCDMA) systems. Numerical results obtained by this study are therefore compared to previous studies employing different codes like two dimensional extended double weight (2D-EDW), two dimensional flexible cross correlation/modified double weight (2D-FCC/MDW), two dimensional perfect difference (2D-PD), two dimensional diluted perfect difference (2D-DPD), two dimensional multi service (2D-MS) and two dimensional zero cross correlation/multi diagonal (2D-ZCC/MD) codes. Accordingly, it is demonstrated that the proposed 2D-CS code outperforms all codes given previously in terms of system capacity where the small increasing percentage is about 40% compared to 2D-ZCC/MD and 2D-MS. Systems using 2D-CS code can support until 203 simultaneous users with a total code length equal to 171. System performance investigation leads to a BER and Q-Factor closely to1.0E?12 and 1.0E?27, and 6.6 dB and 10.6 dB at 20 km of single mode fiber length using white light source and Laser, respectively. Furthermore, such a code can be easily adopted by OCDMA systems for a long distance up to approximately 55 and 100 km.
Local spatio-temporal features have been shown to be effective and robust in order to represent simple actions. However, for high level human activities with long-range motion or multiple interactive body parts and persons, the limitation of low-level features blows up because of their localness. This paper addresses the problem by suggesting a framework that computes mid-level features and takes into account their contextual information.First, we represent human activities by a set of mid-level components, referred to as activity components, which have consistent structure and motion in spatial and temporal domain respectively. These activity components are extracted hierarchically from videos, i.e., extracting key-points, grouping them into trajectories and finally clustering trajectories into components. Second, to further exploit the interdependencies of the activity components, we introduce a spatio-temporal context kernel (STCK), which not only captures local properties of features but also considers their spatial and temporal context information. Experiments conducted on two challenging activity recognition datasets show that the proposed approach outperforms standard spatio-temporal features and our STCK context kernel improves further the performance. 相似文献
Silicon - Silicon (Si) plays an important role in providing beneficial effects on plant growth and yield, especially under stressful environments such as salinity. The objective of this work is to... 相似文献