Abstract
In this work, we propose a robust tracking algorithm based on context-aware correlation filter framework. In order to improve the richness of the feature representation, the proposed hyper-feature which contains linearly weighted mixture of hand-crafted features (such as HOG, color histogram) and hierarchical deep convolutional features (such as VGGNet). The final output response map is optimized by the Gaussian constrained optimization method to control the response map follow the Gaussian distribution, which gain the robustness to target appearance variations. In addition, in terms of model update, an efficient adaptive model updating method is proposed to suppress the model noises significantly. Extensive experimental results on well-known tracking benchmark datasets to evaluate the proposed algorithm. Experimental results demonstrate that the proposed algorithm performs favorably against many state-of-the-art methods in terms of success rate, accuracy, and robustness.
Highlights
Visual tracking is a popular research problem in computer vision community with a wide range of applications such as intelligent video surveillance [1], automatic drive [2], medical diagnosis [3] and virtual reality [4]
These results demonstrates that the effectiveness of our methods regarding hyper-feature fusion method and model update strategy
The proposed perform more robust for scenes with fast motion, motion blur, out-of-view, in-plane-rotation and out-of-rotation with the help of adaptive model update method, and the fusion of hyper-features retain powerful discriminative to target appearance variation
Summary
Visual tracking is a popular research problem in computer vision community with a wide range of applications such as intelligent video surveillance [1], automatic drive [2], medical diagnosis [3] and virtual reality [4]. To better address the above-mentioned issues and take full advantage of the performance between different features, a robust hyper-feature tracking method is proposed, which combines of multiple hierarchical convolutional features and hand-crafted features by linearly weighting. Combining hand-crafted and hierarchical convolutional features as target representation for visual tracking, and propose an adaptive fusion method for taking full advantage of their complementary performance to enhance discriminative ability. Danelljan et al [27] employed deep convolution features based on discriminative correlation filter framework instead of the traditional hand-crafted features It demonstrates that the deep convolution feature achieves superior performance in solving the tracking problem. Song et al [33] proposed a robust convolution residual learning tracking method, which integrates feature extraction, response map generation, and model updating into a layer of convolutional neural network for end-to-end training.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
