Filters








6,432 Hits in 7.0 sec

Learning to Adversarially Blur Visual Object Tracking [article]

Qing Guo and Ziyi Cheng and Felix Juefei-Xu and Lei Ma and Xiaofei Xie and Yang Liu and Jianjun Zhao
2021 arXiv   pre-print
Motion blur caused by the moving of the object or camera during the exposure can be a key challenge for visual object tracking, affecting tracking accuracy significantly.  ...  In this work, we explore the robustness of visual object trackers against motion blur from a new angle, i.e., adversarial blur attack (ABA).  ...  In contrast, the adversarial blur causes a significant performance drop, demonstrating the adversarial blur does pose threat to visual object tracking.  ... 
arXiv:2107.12085v4 fatcat:vrlgbw22rngbtfeer5es36ryrm

Adversarial Feature Sampling Learning for Efficient Visual Tracking [article]

Yingjie Yin, Lei Zhang, De Xu, Xingang Wang
2018 arXiv   pre-print
In this paper, we propose a new visual tracking method using sampling deep convolutional features to address this problem.  ...  In addition, a generative adversarial network is integrated into our network framework to augment positive samples and improve the tracking performance.  ...  We would like to thank NVIDIA for providing the GPU card.  ... 
arXiv:1809.04741v2 fatcat:gmboqsu7ojhnnc2ggtsoumr2hi

SINT++: Robust Visual Tracking via Adversarial Positive Instance Generation

Xiao Wang, Chenglong Li, Bin Luo, Jin Tang
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
In this paper, we propose to generate hard positive samples via adversarial learning for visual tracking.  ...  Existing visual trackers are easily disturbed by occlusion, blur and large deformation.  ...  This may lead to under-fitting of learned classifier for visual tracking. In this paper, we propose to utilize the variational autoencoder (VAE) to learn the target object manifold.  ... 
doi:10.1109/cvpr.2018.00511 dblp:conf/cvpr/WangL0T18 fatcat:ovjvqxtcozawpmwxtikgbblcy4

Sanitization of Visual Multimedia Content: A Survey of Techniques, Attacks, and Future Directions [article]

Hanaa Abbas, Roberto Di Pietro
2022 arXiv   pre-print
This paper presents a review of the mechanisms designed for protecting digital visual contents (i.e., images and videos), the attacks against the cited mechanisms, and possible countermeasures.  ...  Data sanitization -- the process of obfuscating or removing sensitive content related to the data -- helps to mitigate the severe impact of potential security and privacy risks.  ...  This publication was also partially supported by awards NPRP-S-11-0109-180242 "Extending Blockchain Technology -a Novel Paradigm and its Applications to Cybersecurity and Fintech" from the QNRF-Qatar National  ... 
arXiv:2207.02051v1 fatcat:oevtpttxvvgo3p537t54wwhq5y

Localization-Aware Meta Tracker Guided with Adversarial Features

Yiting Lin, Bineng Zhong, Guorong Li, Sicheng Zhao, Ziyi Chen, Wentao Fan
2019 IEEE Access  
INDEX TERMS Adversarial features, localization accuracy, visual object tracking.  ...  Deep learning has recently shown great potentials in learning powerful features for visual tracking.  ...  ADVERSARIAL LEARNING BASED TRACKERS Inspired by the success of adversarial learning [15] , there have been some attempts to enhance the robustness of tracking algorithms using adversarial learning [9  ... 
doi:10.1109/access.2019.2930550 fatcat:clffotdbwjgvnenyrcehfqe3qa

Watch out! Motion is Blurring the Vision of Your Deep Neural Networks [article]

Qing Guo and Felix Juefei-Xu and Xiaofei Xie and Lei Ma and Jian Wang and Bing Yu and Wei Feng and Yang Liu
2020 arXiv   pre-print
We propose a novel adversarial attack method that can generate visually natural motion-blurred adversarial examples, named motion-based adversarial blur attack (ABBA).  ...  To generate visually more natural and plausible examples, we further propose the saliency-regularized adversarial kernel prediction, where the salient region serves as a moving object, and the predicted  ...  We will also discuss the effects of motion blur to visual object tracking [41, 42, 43] , through the proposed ABBA and recent attacking method [44] against tracking.  ... 
arXiv:2002.03500v3 fatcat:tsfdhjbhiff5fno3qgiq4usymy

Robust Visual Tracking Based on Hybrid Network and Similarity Grouping

Yu Liu, Xiaoqiang Li, Chen Huang, Dian-hua Zhang, Ming-ke Gao
2019 Australian Journal of Intelligent Information Processing Systems  
The proposed HNM adopts recurrent neural networks (RNNs) to model the self-structure of object, and utilizes adversarial learning to enhance the representation ability of the most robust features in temporal  ...  In addition, in order to track the target according to its categoryspecific information, we propose a category-specific similarity grouping algorithm so as to activate the corresponding single branch from  ...  To solve the problem of lack of sufficient training data specialized for visual tracking, H.Nam [10] proposed a Multi-Domain Network (MDNet) based on CNNs to learn a unified representation of objects  ... 
dblp:journals/ajiips/LiuLHZG19 fatcat:ivqf7unkefeuzj5xy3fduglexe

Tracking by Joint Local and Global Search: A Target-aware Attention based Approach [article]

Xiao Wang, Jin Tang, Bin Luo, Yaowei Wang, Yonghong Tian, Feng Wu
2021 arXiv   pre-print
Tracking-by-detection is a very popular framework for single object tracking which attempts to search the target object within a local search window for each frame.  ...  More importantly, we resort to adversarial training for better attention prediction.  ...  Effects of Appearance Adversarial Learning. To obtain better attention maps, we introduce the adversarial learning strategy to model the relations between different pixel values.  ... 
arXiv:2106.04840v1 fatcat:3pktzebc6vfzrmr7ueu7l7qpd4

Why is the video analytics accuracy fluctuating, and what can we do about it? [article]

Sibendu Paul, Kunal Rao, Giuseppe Coviello, Murugan Sankaradas, Oliver Po, Y. Charlie Hu, Srimat Chakradhar
2022 arXiv   pre-print
In particular, we show that our newly trained Yolov5 model reduces fluctuation in object detection across frames, which leads to better tracking of objects(40% fewer mistakes in tracking).  ...  To address this inadvertent adversarial effect from the camera, we explore the use of transfer learning techniques to improve learning in video analytics tasks through the transfer of knowledge from learning  ...  To mitigate this adversarial effect, we propose a transfer learning based approach and train a new Yolov5 model for object detection.  ... 
arXiv:2208.12644v2 fatcat:zfjr2gdkjrcqrbocyypm6cmb7y

Table of contents

2021 IEEE Transactions on Image Processing  
Pan Exploring the Effects of Blur and Deblurring to Visual Object Tracking .  ...  Ding Learning Diverse Models for End-to-End Ensemble Tracking ............................. N. Wang, W.  ...  Xu 3D Object Representation Learning: A Set-to-Set Matching Perspective ....... T. Yu, J. Meng, M. Yang, and J.  ... 
doi:10.1109/tip.2021.3129401 fatcat:jfcp3vj26vgahpbufknwbg4y2i

Towards More Flexible and Accurate Object Tracking with Natural Language: Algorithms and Benchmark [article]

Xiao Wang, Xiujun Shu, Zhipeng Zhang, Bo Jiang, Yaowei Wang, Yonghong Tian, Feng Wu
2021 arXiv   pre-print
We also introduce two new challenges into TNL2K for the object tracking task, i.e., adversarial samples and modality switch.  ...  In this work, we propose a new benchmark specifically dedicated to the tracking-by-language, including a large scale dataset, strong and diverse baseline methods.  ...  Besides, these benchmarks also ignore the adversarial samples which limit the development of adversarial learning-based trackers [30, 45, 67, 73] .  ... 
arXiv:2103.16746v1 fatcat:hxmyyxiaenhyjhyitwiazjkyea

Hiding Behind Backdoors: Self-Obfuscation Against Generative Models [article]

Siddhartha Datta, Nigel Shadbolt
2022 arXiv   pre-print
Attack vectors that compromise machine learning pipelines in the physical world have been demonstrated in recent research, from perturbations to architectural components.  ...  Our contribution is to describe, implement and evaluate a generalized attack, in the hope of raising awareness regarding the challenge of architectural robustness within the machine learning community.  ...  Object detection is a preprocessing component prior to person re-identification/tracking; adversarially attacking this component given prior knowledge (e.g. trained on MS COCO) allows evasion of detection  ... 
arXiv:2201.09774v1 fatcat:iexoeg3rajb7nmjvqu3hqtttzi

Adversarial Spatio-Temporal Learning for Video Deblurring

Kaihao Zhang, Yongzhen Huang, Yong Du, Liang Wang
2018 IEEE Transactions on Image Processing  
Camera shake or target movement often leads to undesired blur effects in videos captured by a hand-held camera.  ...  In this paper, to address the first challenge, we propose a DeBLuRring Network (DBLRNet) for spatial-temporal learning by applying a modified 3D convolution to both spatial and temporal domains.  ...  The KITTI dataset consists of several subsets for various kinds of tasks, such as stereo matching, optical flow estimation, visual odometry, 3D object detection and tracking.  ... 
doi:10.1109/tip.2018.2867733 pmid:30176588 fatcat:c2q7hrbh6ncuralkv32b2hqjne

Underwater Image Enhancement Using Deep Residual Framework

Prof. Anuja Phapale, Atal Deshmukh, Keshav Katkar, Onkar Karale, Puja Kasture
2021 International Journal of Scientific Research in Science and Technology  
traditional methods & deep learning models.  ...  Firstly, the generation of synthetic underwater images takes place for which cycle-consistent adversarial networks (CycleGAN) is employed.  ...  Forward scattering light usually contributes to the blurred texture features of underwater objects as it comes from the object.  ... 
doi:10.32628/ijsrst218314 fatcat:x3vq4jnpgve2njoqm2lq2dkovq

PAMSGAN: Pyramid Attention Mechanism-oriented Symmetry Generative Adversarial Network for Motion Image Deblurring

Zhenfeng Zhang
2021 IEEE Access  
Therefore, the PAMSGAN achieves good results in subjective visual effects and objective evaluation results.  ...  Research on the restoration of blurred images will help to improve the visual quality of images from the perspective of human vision [3] .  ... 
doi:10.1109/access.2021.3099803 fatcat:2jwholnombfx7ebcmojuyzim7q
« Previous Showing results 1 — 15 out of 6,432 results