Spatially Attentive Visual Tracking Using Multi-Model Adaptive Response Fusion

Abstract

Recent years have witnessed the top performances of integrating multi-level features from the pre-trained convolutional neural network (CNN) into correlation filters framework. However, they still suffer from background interference in detection stage due to the large search region and contamination of training samples caused by inaccurate tracking. In this paper, to suppress the interference of background features in target detection stage, an effective spatial attention map (SAM) is proposed to differently weight the multi-hierarchical convolutional features from search region to obtain the attentional features. This way helps to reduce the filter values corresponding to background features. Moreover, we construct multiple elementary correlation filter (ECF) models on multi-hierarchical deep features from CNN to track the target in parallel. To further improve the tracking stability, a multi-model adaptive response fusion (MAF) mechanism is presented. The mechanism can adaptively choose the outputs of reliable ECF models for adaptive weighted fusion by evaluating the confidences of response maps generated by attentional features convolved with ECF models. Finally, to adapt the target appearance changes in the following frames and avoid model corruption, we propose an adaptive updating strategy for the updates of the SAM and ECF models. We perform comprehensive experiments on OTB-2013 and OTB-2015 datasets and the experimental results show the superiority of our algorithm over other 12 state-of-the-art approaches.INDEX TERMS Adaptive updating strategy, convolutional neural network, correlation filter, multi-model adaptive response fusion mechanism, object tracking, spatial attention map.

Adaptive Leader-Follower Formation Control of Under-actuated Surface Vessels with Model Uncertainties and Input Constraints

“…The control objective is to make the follower ASVs track the leaders with some prescribed offsets. In recent years, many visual object tracking techniques based on correlation filters [8,9] and deep learning [10,11] have emerged. Other advantages of the leader-follower formation are simplicity, understandability and its easy implementation.…”

Effect of Load Priority Modeling on the Size of Fuel Cell as an Emergency Power Unit in a More-Electric Aircraft

“…Now in the era of artificial intelligence, deep learning has achieved great success [2][3][4], because it requires very little engineering by hand, so it can easily use the increase of available computation and data. In order to monitor and track [5][6][7] the surrounding objects in real time, the intelligent algorithm used in aircraft will consume a lot of power. Intelligent aircraft needs stable and reliable electric supply, and the electric consumption is increasing.…”

“…b C b,q,t and b D b,q,t are the binary variables and indicate the charging and discharging status of battery b connected at bus q and at time t. Also, SoC b,q,t in (6) indicates the state of charge (SoC) battery b connected at bus q and at time t which is limited by its boundaries. The dynamic model of energy exchange in the battery is shown in (7). In this regard, η C b and η D b are the charging and the discharging efficiency of battery b, respectively.…”

A bias‐compensated proportionate NLMS algorithm with noisy input signals

“…Conventional adaptive filters are usually assumed as a linear system, and there are many research topics related to a variable step‐size strategy, regularization parameter adjustment, sparse system, and various noise scenarios. Moreover, nonlinear filters such as Volterra filters and neural network–based filters have been researched actively due to their practical advantages in adaptive signal processing . Among them, the least‐mean‐squares (LMS) algorithm and normalized LMS (NLMS) algorithm have been well‐known adaptive filtering algorithms because of their low computational complexity and ease of implementation.…”

“…Moreover, nonlinear filters such as Volterra filters and neural network-based filters have been researched actively due to their practical advantages in adaptive signal processing. [6][7][8][9] Among them, the least-mean-squares (LMS) algorithm and normalized LMS (NLMS) algorithm have been well-known adaptive filtering algorithms because of their low computational complexity and ease of implementation.…”