Filters








94,311 Hits in 7.9 sec

Visual Object Tracking by Using Ranking Loss

Hakan Cevikalp, Hasan Saribas, Burak Benligiray, Sinem Kahvecioglu
2019 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)  
To this end, we employ a ranking loss which provides a fine-tuning of the target object position and returns more precise bounding boxes framing the target object.  ...  When the proposed network is used with a simple yet effective model update rule, our proposed tracker achieves the state-of-the-art results on all tested challenging tracking datasets.  ...  The authors also would like to thank NVIDIA for GPU donation used in this study.  ... 
doi:10.1109/iccvw.2019.00280 dblp:conf/iccvw/CevikalpSBK19 fatcat:tscwm77d2fbt3hdft7kz7yz2em

Target-Aware Deep Tracking [article]

Xin Li, Chao Ma, Baoyuan Wu, Zhenyu He, Ming-Hsuan Yang
2019 arXiv   pre-print
Despite demonstrated successes for numerous vision tasks, the contributions of using pre-trained deep features for visual tracking are not as significant as that for object recognition.  ...  The key issue is that in visual tracking the targets of interest can be arbitrary object class with arbitrary forms.  ...  Given a target object specified by a bounding box in the first frame, visual tracking aims to locate the target object in the subsequent frames.  ... 
arXiv:1904.01772v1 fatcat:ttqdj7sc4jaxxanbaamsuiv6ye

Target-Aware Deep Tracking

Xin Li, Chao Ma, Baoyuan Wu, Zhenyu He, Ming-Hsuan Yang
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Despite demonstrated successes for numerous vision tasks, the contributions of using pre-trained deep features for visual tracking are not as significant as that for object recognition.  ...  The key issue is that in visual tracking the targets of interest can be arbitrary object class with arbitrary forms.  ...  Given a target object specified by a bounding box in the first frame, visual tracking aims to locate the target object in the subsequent frames.  ... 
doi:10.1109/cvpr.2019.00146 dblp:conf/cvpr/Li0WH019 fatcat:hjod2ekhzbfjdonfhfxi6iekha

Comprehensive Underwater Object Tracking Benchmark Dataset and Underwater Image Enhancement With GAN

Karen Panetta, Landry Kezebou, Victor Oludare, Sos Agaian
2021 IEEE Journal of Oceanic Engineering  
Inherent underwater distortions, such as color loss, poor contrast, and underexposure, caused by attenuation of light, refraction, and scattering, greatly affect the visual quality of underwater data,  ...  We also evaluate the visual quality of our model's output against existing GAN-based methods using well-accepted quality metrics and demonstrate that our model yields better visual data.  ...  The recent success in object tracking has been facilitated by dedicated benchmarking datasets such as object tracking benchmark (OTB) [5] , [6] , visual object tracking (VOT) [7] , and multiple object  ... 
doi:10.1109/joe.2021.3086907 fatcat:sspilleihrepjhnkt4clzw3jci

Ranking-Based Siamese Visual Tracking [article]

Feng Tang, Qiang Ling
2022 arXiv   pre-print
Current Siamese-based trackers mainly formulate the visual tracking into two independent subtasks, including classification and localization.  ...  fooled by the distractors.  ...  Introduction Visual object tracking aims to estimate the location information of an arbitrary target in each frame of a video sequence.  ... 
arXiv:2205.11761v1 fatcat:rnq57xv4t5egta5kbrdpbkib3i

Robust visual tracking via samples ranking

Heyan Zhu, Hui Wang
2019 EURASIP Journal on Advances in Signal Processing  
In recent years, deep convolutional neural networks (CNNs) have achieved great success in visual tracking.  ...  This is especially crucial for visual tracking because there is only one best target candidate among all positive candidates, which tightly bounds the target.  ...  Such success can be attributed to powerful deep CNN features regularized by both the classification loss and the ranking loss, as well as the spatial location loss. Camera motion.  ... 
doi:10.1186/s13634-019-0639-z fatcat:sj6p4pl6sbdk5mtfdklgirm7h4

Quadruplet Network with One-Shot Learning for Fast Visual Object Tracking [article]

Xingping Dong and Jianbing Shen and Yu Liu and Wenguan Wang and Fatih Porikli
2018 arXiv   pre-print
We evaluate our quadruplet framework by model-free tracking-by-detection of objects from a single initial exemplar in several Visual Object Tracking benchmarks.  ...  We design four shared networks that receive multi-tuple of instances as inputs and are connected by a novel loss function consisting of pair-loss and triplet-loss.  ...  Results on VOT-2015 In our evaluations, we use the Visual Object Tracking 2015 (VOT-2015) toolkit, which contains the evaluation in short-term visual object tacking tasks.  ... 
arXiv:1705.07222v2 fatcat:toqmnji66zhm5alzbksjlveu74

A Ranking Based Attention Approach for Visual Tracking

Shenhui Peng, Sei-ichiro Kamata, Toby P. Breckon
2019 2019 IEEE International Conference on Image Processing (ICIP)  
In visual tracking algorithms, the object categories should not be limited by the training set.  ...  Since the specific kind of target will not be known in advance of the tracking task, the visual tracking algorithm should be robust enough to track any kind of the object and can be promptly specialized  ... 
doi:10.1109/icip.2019.8803358 dblp:conf/icip/PengKB19 fatcat:gla5cbftwjakzllblfr5it6vru

Proposal-Based Visual Tracking Using Spatial Cascaded Transformed Region Proposal Network

Ximing Zhang, Shujuan Luo, Xuewu Fan
2020 Sensors  
We extensively prove the effectiveness of our proposed method on the ablation studies of the tracking datasets, which include OTB-2015 (Object Tracking Benchmark 2015), VOT-2018 (Visual Object Tracking  ...  2018), LaSOT (Large Scale Single Object Tracking), TrackingNet (A Large-Scale Dataset and Benchmark for Object Tracking in the Wild) and UAV123 (UAV Tracking Dataset).  ...  Acknowledgments: Thanks to the experimental data provided by University of Ljubljana, SICK, Hiar, King Abdullah University of Science and Technology.  ... 
doi:10.3390/s20174810 pmid:32858907 pmcid:PMC7506765 fatcat:2kuuqlhvmzdozfddxq7ox6vql4

TIED: A Cycle Consistent Encoder-Decoder Model for Text-to-Image Retrieval

Clint Sebastian, Raffaele Imbriaco, Panagiotis Meletis, Gijs Dubbelman, Egor Bondarev, Peter H.N. de With
2021 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
The method exploits visual semantic attributes of a target vehicle along with a cycle-consistency loss.  ...  Retrieving specific vehicle tracks by Natural Language (NL)-based descriptions is a convenient way to monitor vehicle movement patterns and traffic-related events.  ...  Contrary to NL-based image or object retrieval [14, 11] , an NL-based track retrieval system should address the time dimension of the task, as indicated by the related NL-based visual object tracking  ... 
doi:10.1109/cvprw53098.2021.00467 fatcat:ooikmrmmxrbxhiy4lshfv3o3gi

Tracking the Untrackable [article]

Fangyi Zhang
2020 arXiv   pre-print
Although short-term fully occlusion happens rare in visual object tracking, most trackers will fail under these circumstances.  ...  Inspired by that, we present a HAllucinating Features to Track (HAFT) model that enables to forecast the visual feature embedding of future frames.  ...  Each sequence is per-frame annotated by five visual attributes, and the bounding box is generated from pixel-wise segmentation of the tracked object.  ... 
arXiv:2007.10148v1 fatcat:kdeky4xe3jem5jvvzlcrznfcse

Unsupervised Learning of Visual Representations using Videos [article]

Xiaolong Wang, Abhinav Gupta
2015 arXiv   pre-print
That is, two patches connected by a track should have similar visual representation in deep feature space since they probably belong to the same object or object part.  ...  We design a Siamese-triplet network with a ranking loss function to train this CNN representation.  ...  Acknowledgement: This work was partially supported by ONR MURI N000141010934 and NSF IIS 1320083.  ... 
arXiv:1505.00687v2 fatcat:q4s5vgctl5ejpe6mo63iobgbxu

A CBIR-based evaluation framework for visual attention models

Dounia Awad, Matei Mancas, Nicolas Riche, Vincent Courboulay, Arnaud Revel
2015 2015 23rd European Signal Processing Conference (EUSIPCO)  
The computational models of visual attention, originally proposed as cognitive models of human attention, nowadays are being used as front-ends to numerous vision systems like automatic object recognition  ...  These findings suggest that the saliency models ranking might be different for each application and the use of eye-tracking rankings to choose a model for a given application is not optimal.  ...  In [3] it is already shown that depending on the ground truth (eye-tracking data or manually segmented objects), the saliency models ranking can be very different.  ... 
doi:10.1109/eusipco.2015.7362639 dblp:conf/eusipco/AwadMRCR15 fatcat:6kyfdkevhnfgndkagpsuf6v3gy

Unsupervised Learning of Visual Representations Using Videos

Xiaolong Wang, Abhinav Gupta
2015 2015 IEEE International Conference on Computer Vision (ICCV)  
That is, two patches connected by a track should have similar visual representation in deep feature space since they probably belong to the same object or object part.  ...  We design a Siamese-triplet network with a ranking loss function to train this CNN representation.  ...  Acknowledgement: This work was partially supported by ONR MURI N000141010934 and NSF IIS 1320083.  ... 
doi:10.1109/iccv.2015.320 dblp:conf/iccv/WangG15 fatcat:kqxttowrq5f75kn34je3vdjc2i

Siamese Visual Tracking with Residual Fusion Learning

Xinglong Sun, Guangliang Han, Lihong Guo
2021 IEEE Access  
Specifically, the network employs the deep-layer features as direct input to semantically recognize the object from background, and refines the object state with local detail patterns by exploring the  ...  Multi-stage feature fusion is pretty effective for deep Siamese trackers to promote tracking performance.  ...  Sun et al.: Siamese Visual Tracking with Residual Fusion Learningbetween two objects with the proposed fusion framework.V.  ... 
doi:10.1109/access.2021.3134066 fatcat:b3xdu2bumjakdd7m6tglsiripu
« Previous Showing results 1 — 15 out of 94,311 results