Robust Visual Tracking based on Adversarial Unlabeled Instance Generation with Label Smoothing Loss Regularization

Han, Yamin; Zhang, Peng; Zha, Yufei; Cooper, Garth Douglas; Zhang, Yanning

doi:10.1016/j.patcog.2019.107027

Cited by 24 publications

(13 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GANs have been utilized in other image processing and computer vision tasks [354]- [357] such as object transfiguration [358], [359], semantic segmentation [360], visual saliency prediction [361], object tracking [362], [363], image dehazing [364]- [366], natural image matting [367], image inpainting [368], [369], image fusion [370], image completion [371], and image classification [372].…”

Section: Other Image and Vision Applicationsmentioning

confidence: 99%

A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications

Gui

Sun

Wen

et al. 2023

IEEE Trans. Knowl. Data Eng.

583

224

View full text Add to dashboard Cite

Generative adversarial networks (GANs) are a hot research topic recently. GANs have been widely studied since 2014, and a large number of algorithms have been proposed. However, there is few comprehensive study explaining the connections among different GANs variants, and how they have evolved. In this paper, we attempt to provide a review on various GANs methods from the perspectives of algorithms, theory, and applications. Firstly, the motivations, mathematical representations, and structure of most GANs algorithms are introduced in details. Furthermore, GANs have been combined with other machine learning algorithms for specific applications, such as semi-supervised learning, transfer learning, and reinforcement learning. This paper compares the commonalities and differences of these GANs methods. Secondly, theoretical issues related to GANs are investigated. Thirdly, typical applications of GANs in image processing and computer vision, natural language processing, music, speech and audio, medical field, and data science are illustrated. Finally, the future open research problems for GANs are pointed out.

show abstract

Section: Other Image and Vision Applicationsmentioning

confidence: 99%

A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications

Gui

Sun

Wen

et al. 2023

IEEE Trans. Knowl. Data Eng.

583

224

View full text Add to dashboard Cite

show abstract

“…Occlusion completely, out of sight, similar interference UPDT [35] e wide Gaussian label function expands the area, rotates the blur operation, and captures semantic information Occlusion completely, background clutter, out of sight, scale variations SINT++ [36] Samples were collected by traversing the target manifold structure to generate objects that did not appear in the training data Partial cover, deformation, elapsed time Reference [37] Generate adversarial network, and generate deformation, fuzzy samples, adaptive learning discrimination features Deformation, motion blur, partial cover and increase the tracking time in the complex scene covered.…”

Section: Adaptive Tracking Algorithm Model Based On Activementioning

confidence: 99%

Analysis of the Occlusion Interference Problem in Target Tracking

Zhang

Zheng

2022

Mathematical Problems in Engineering

View full text Add to dashboard Cite

As an indispensable part in the field of computer vision, target tracking has been widely used in intelligent transportation, missile guidance, unmanned aerial vehicle (UAV) tracking, and many other fields. It has become one of the hot directions in computer vision in recent years, while occlusion problem has always been a great difficulty and challenge in the process of target tracking. In this article, the problem of occlusion interference in target tracking is described, and the solution of occlusion problem is proposed based on different occlusion conditions. Due to the disadvantages of feature point center weighting, multiparticle template matching, and Kalman filter trajectory prediction algorithms in different cases, some algorithms with higher robustness and stability are developed to solve the occlusion problem. In the analysis of the anti-occlusion model, it is found that some tracking errors caused by occlusion can be solved by improving the quality of negative training samples and enriching the diversity of positive sample sets. According to the different training characteristics of online and offline tracking algorithms, the anti-occlusion model suitable for an active learning algorithm under different tracking conditions is found, and the tracking algorithm and characteristics of the active learning algorithm are listed, which is helpful to select the suitable tracking model in different scenarios. Finally, the future development of occlusion problem in target tracking is prospected.

show abstract

“…Han et al. [13] learn occlusion masks to make the classification layers more robust to object occlusion. The mask has binary values, where 0 indicates the presence of occlusion and 1 indicates no occlusion.…”

Section: Related Workmentioning

confidence: 99%

Object tracking using temporally matching filters

Robeson

Javanmardi

2021

IET Computer Vision

View full text Add to dashboard Cite

One of the primary challenges of visual tracking is the variable appearance of the target object. As tracking proceeds, the target object can change its appearance due to illumination changes, rotations, deformations etc. Modern trackers incorporate online updating to learn how the target changes over time. However, they do not use the history of target appearance. To address this shortcoming, we uniquely use domain adaptation with the target appearance history to efficiently learn a temporally matching filter (TMF) during online updating. This TMF emphasizes the persistent features found in different appearances of the target object. It also improves the classification accuracy of the convolutional neural network by assisting the training of the classification layers without incurring the runtime overhead of updating the convolutional layers. Extensive experimental results demonstrate that the proposed TMF-based tracker, which incorporates domain adaptation with the target appearance history, improves tracking performance on three benchmark video databases (OTB-50, OTB-100 and VOT2016) over other online learning trackers. Specifically, it improves the overlap success of VITAL and MDNet by 0.44 % and 1.03 % on the OTB-100 dataset and improves the accuracy of VITAL and MDNet by 0.55 % and 0.06 % on the VOT2016 dataset, respectively.This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

show abstract

Robust Visual Tracking based on Adversarial Unlabeled Instance Generation with Label Smoothing Loss Regularization

Cited by 24 publications

References 15 publications

A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications

A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications

Analysis of the Occlusion Interference Problem in Target Tracking

Object tracking using temporally matching filters

Contact Info

Product

Resources

About