The method based on the two-stream networks has achieved great success in video action recognition. However, most existing methods employ the same structure for both spatial and temporal networks, leading to unsatisfied performance. In this paper, we propose a spatiotemporal heterogeneous two-stream network, which employs two different network structures for spatial and temporal information, respectively. Specifically, the Residual network (ResNet) and BN-Inception are utilized as the base networks to present the spatiotemporal characteristics of different human actions. In addition, a segmental architecture is employed to model long-range temporal structure over video sequences to better distinguish the similar actions owning sub-action sharing phenomenon. Moreover, combined with the strategy of data augment, a modified cross-modal pre-training strategy is proposed and applied to the spatiotemporal heterogeneous network to improve the final performance of human actions recognition. The experiments on UCF101 and HMDB51 datasets demonstrate the proposed spatiotemporal heterogeneous two-stream network outperforms the spatiotemporal isomorphic networks and other related methods.INDEX TERMS Action recognition, spatiotemporal heterogeneous, two-stream networks, ResNet, longrange temporal structure, training strategies.
: Location information is a key issue for applications of the Internet of Things. In this paper, we focus on mobile wireless networks with moving agents and targets. The positioning process is divided into two phases based on the factor graph, i.e., a prediction phase and a joint self-location and tracking phase. In the prediction phase, we develop an adaptive prediction model by exploiting the correlation of trajectories within a short period to formulate the prediction message. In the joint positioning phase, agents calculate the cooperative messages according to variational message passing and locate themselves. Simultaneously, the average consensus algorithm is employed to realize distributed target tracking. The simulation results show that the proposed prediction model is adaptive to the random movement of nodes. The performance of the proposed joint self-location and tracking algorithm is better than the separate cooperative self-localization and tracking algorithms.
<p>A new hybrid algorithm named EODT-LS-SVR based on least squares support vector regression (LS-SVR) with wavelet-based EODT algorithm as preprocessed tools is proposed for removing the interferences and developing the quantitative models with high precision in near-infrared (NIR) spectra. EODT-LS-SVR algorithm is composed of two steps. In the first step, the preprocessing algorithm named EODT, which combines the ideas of wavelet packet transform (WPT), orthogonal signal correction (OSC) and information theory, is employed for the characteristic extraction of analyte information through multi-scale analysis. Entropy-based baseline signal removing (EBSR) algorithm is applied to remove the baseline of the spectra based on information theory with WPT-based analysis, and then the information orthogonal to the concentrations of analyte is removed by OSC algorithm in each frequency band of spectra. In the second step, LS-SVR method coupled with grid search and particle swarm optimization (PSO) technique for parameters optimization is used to enhancing the quality of regression models. EODT-LS-SVR algorithm was validated by two NIR spectral datasets, one used for measuring the fat concentration of milk and the other used for measuring the oil content of corn. The comparison of prediction results demonstrated that the performance of calibration models developed by EODT-LS-SVR algorithm is better than that developed by other conventional algorithms, showing the high efficiency and the high quality for quantitative model development in NIR spectra of complex samples.</p>
Due to the important relationship between ultrasonic velocity and some properties of sample, the measurement of velocity of ultrasound is widely needed in various fields. In this paper, based on phase-locked loop technique, a new hybrid circuit is designed for ultrasonic velocity measurement and named as UV-PLL. In order to improve the stability of phase and achieve the ability of fast locking, an auxiliary capturing circuit, which consists of phase shift circuit and capturing circuit, is designed and implemented in this module. Additionally, two methods for estimating the propagation delay is compared and described in detail. The UV-PLL module is validated through the ultrasonic velocity measurement in distilled water at different temperatures. Experimental results show that the maximum relative deviation of velocity measurement is less than 0.15% and the loop can be quickly locked, indicating that this module can meet the requirements of online measurement and can be widely applied.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.