Filters








259,946 Hits in 6.4 sec

Multiple Object Tracking with Correlation Learning [article]

Qiang Wang, Yun Zheng, Pan Pan, Yinghui Xu
2021 arXiv   pre-print
Recent works have shown that convolutional networks have substantially improved the performance of multiple object tracking by simultaneously learning detection and appearance features.  ...  With extensive experimental results on the MOT datasets, our approach demonstrates the effectiveness of correlation learning with the superior performance and obtains state-of-the-art MOTA of 76.5% and  ...  Our pyramid correlation leverages the natural spatial-temporal coherence in videos. Multi-object tracking can be decomposed into multiple independent single-object tracking.  ... 
arXiv:2104.03541v1 fatcat:2bektgwrorbwbi72ll4i4wyc2m

Recent Advances in Embedding Methods for Multi-Object Tracking: A Survey [article]

Gaoang Wang, Mingli Song, Jenq-Neng Hwang
2022 arXiv   pre-print
Multi-object tracking (MOT) aims to associate target objects across video frames in order to obtain entire moving trajectories.  ...  cross-frame joint embedding, correlation embedding, sequential embedding, tracklet embedding, and cross-track relational embedding.  ...  CLEAR MOT [233] measures the multiple object tracking accuracy (MOTA) and multiple object tracking precision (MOPT) between the detected boxes and ground truth boxes.  ... 
arXiv:2205.10766v1 fatcat:p7s7lnnlsnadrhsdcmwlg7msfy

A Robust Visual Tracking Method through Deep Learning Features

Jia-zhen XU, Ming-zhang ZUO, Lin YANG, Lei HUANG
2017 DEStech Transactions on Computer Science and Engineering  
Object tracking is one of the most important components in many applications of computer vision.  ...  In this paper, we propose a novel approach based on correlation filter framework for robust scale estimation through deep learning features.  ...  For tracking task, we need to learn instance rather than categories because it is not unusual when multiple instances of same category appears in the same frame.  ... 
doi:10.12783/dtcse/aita2016/7562 fatcat:xhhecaysvngyzitndm27u4lfei

Learning Local–Global Multiple Correlation Filters for Robust Visual Tracking with Kalman Filter Redetection

Jianming Zhang, Yang Liu, Hehua Liu, Jin Wang
2021 Sensors  
In this paper, we propose a local–global multiple correlation filters (LGCF) tracking algorithm for edge computing systems capturing moving targets, such as vehicles and pedestrians.  ...  Then, we propose a local–global collaborative strategy to exchange information between local and global correlation filters. This strategy can avoid the wrong learning of the object appearance model.  ...  Based on the local-global learning model, we propose a collaborative update strategy with multiple correlation filters.  ... 
doi:10.3390/s21041129 pmid:33562878 fatcat:2yxnngieu5huxjmpaep4vp2sd4

Learning Multi-task Correlation Particle Filters for Visual Tracking

Tianzhu Zhang, Changsheng Xu, Ming-Hsuan Yang
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Compared with existing tracking methods based on correlation filters and particle filters, the proposed MCPF enjoys several merits.  ...  We first present the multi-task correlation filter (MCF) that takes the interdependencies among different object parts and features into account to learn the correlation filters jointly.  ...  We learn the correlation filters for all parts among multiple features jointly.  ... 
doi:10.1109/tpami.2018.2797062 pmid:29994598 fatcat:chpzxggpubgj3opd3dwic55w74

Hierarchical Convolutional Features for Visual Tracking

Chao Ma, Jia-Bin Huang, Xiaokang Yang, Ming-Hsuan Yang
2015 2015 IEEE International Conference on Computer Vision (ICCV)  
Specifically, we adaptively learn correlation filters on each convolutional layer to encode the target appearance. We hierarchically infer the maximum response of each layer to locate targets.  ...  Visual object tracking is challenging as target objects often undergo significant appearance changes caused by deformation, abrupt motion, background clutter and occlusion.  ...  [34] combine multiple classifiers with different learning rates. On the other hand, Hare et al.  ... 
doi:10.1109/iccv.2015.352 dblp:conf/iccv/MaHYY15 fatcat:gd2v4prsmvh75n5wk5536hubta

Robust Visual Tracking via Hierarchical Convolutional Features [article]

Chao Ma, Jia-Bin Huang, Xiaokang Yang, Ming-Hsuan Yang
2018 arXiv   pre-print
To further handle the issues with scale estimation and re-detecting target objects from tracking failures caused by heavy occlusion or out-of-the-view movement, we conservatively learn another correlation  ...  Deep neural networks trained on object recognition datasets consist of multiple convolutional layers. These layers encode target appearance with different levels of abstraction.  ...  In contrast to the FCNT [62] tracker which uses two convolutional layers for visual tracking, we exploit the feature hierarchies of deep networks by learning adaptive correlation filters over multiple  ... 
arXiv:1707.03816v2 fatcat:rqxytlu64na4hmwtjcahk5otou

Dual Model Learning Combined with Multiple Feature Selection for Accurate Visual Tracking

Jianming Zhang, Xiaokang Jin, Juan Sun, Jin Wang, Keqin Li
2019 IEEE Access  
INDEX TERMS Convolutional neural network, correlation filter, learning models, multiple feature selection, object tracking. 43956 2169-3536  ...  In this paper, we propose dual model learning combined with multiple feature selection for accurate visual tracking.  ...  RELATED WORKS In this section, we mainly introduce three categories of trackers closely related to our algorithm: tracking by deep learning, tracking by correlation filters and tracking by multiple models  ... 
doi:10.1109/access.2019.2908668 fatcat:3k3y3zsu4zda7ajjwf2dfx25z4

Robust Visual Tracking via Hierarchical Convolutional Features

Chao Ma, Jia-Bin Huang, Xiaokang Yang, Ming-Hsuan Yang
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
To further handle the issues with scale estimation and target re-detection from tracking failures caused by heavy occlusion or moving out of the view, we conservatively learn another correlation filter  ...  Deep neural networks trained on object recognition datasets consist of multiple convolutional layers. These layers encode target appearance with different levels of abstraction.  ...  [45] combine multiple classifiers with different learning rates for visual tracking. By alleviating the sampling ambiguity problem, these methods perform well in a recent benchmark study [33] .  ... 
doi:10.1109/tpami.2018.2865311 pmid:30106709 fatcat:dgpu2z5agrcc5mtxtmhvjaxtgu

Multiple Cues-Based Robust Visual Object Tracking Method

Baber Khan, Abdul Jalil, Ahmad Ali, Khaled Alkhaledi, Khizer Mehmood, Khalid Mehmood Cheema, Maria Murad, Hanan Tariq, Ahmed M. El-Sherbeeny
2022 Electronics  
Visual object tracking is still considered a challenging task in computer vision research society.  ...  A comparison of the proposed tracking scheme with other state-of-the-art methods is also presented in this paper.  ...  Occlusion Handling Mechanism The correlation response map gives multiple cues about the target in visual object tracking.  ... 
doi:10.3390/electronics11030345 fatcat:setvps6ah5fdznryefselbzemu

Correlation Tracking via Self-Adaptive Fusion of Multiple Features

Zhi Chen, Peizhong Liu, Yongzhao Du, Yanmin Luo, Wancheng Zhang
2018 Information  
Correlation filter (CF) based tracking algorithms have shown excellent performance in comparison to most state-of-the-art algorithms on the object tracking benchmark (OTB).  ...  In order to address problems that are mentioned above, in this paper, we propose a robust multi-scale correlation filter tracking algorithm via self-adaptive fusion of multiple features.  ...  The Context-Aware Correlation Filter (CACF) framework that is proposed by [20] that integrates surrounding context information into the learned filter, in order to learn a filter with a high response  ... 
doi:10.3390/info9100241 fatcat:3mujgsm5nvfengaggoqzgg4eca

Attention Modulated Multiple Object Tracking with Motion Enhancement and Dual Correlation

Yifeng Wang, Zhijiang Zhang, Ning Zhang, Dan Zeng
2021 Symmetry  
To alleviate the above limitations, we propose a one-shot network named Motion and Correlation-Multiple Object Tracking (MAC-MOT).  ...  Our proposed approach is evaluated on the popular multiple object tracking benchmarks MOT16 and MOT17.  ...  Introduction Multiple object tracking (MOT) aims at estimating the locations of multiple objects and providing identifications.  ... 
doi:10.3390/sym13020266 fatcat:c35a3dao2zfm3eqntdnpk7xlzi

Unsupervised Deep Tracking [article]

Ning Wang, Yibing Song, Chao Ma, Wengang Zhou, Wei Liu, Houqiang Li
2019 arXiv   pre-print
Meanwhile, we propose a multiple-frame validation method and a cost-sensitive loss to facilitate unsupervised learning.  ...  We propose an unsupervised visual tracking method in this paper.  ...  Second, our unsupervised framework is coupled with a tracking objective function, so the learned feature representation is effective in presenting the generic target objects.  ... 
arXiv:1904.01828v1 fatcat:dmsqfi7xqzayhg3k26pmb4bbhu

FAST AND ROBUST MODEL FOR MULTIPLE OBJECTS TRACKING USING KEY-FRAME DETECTION AND CO-TRAINED CLASSIFIER

Phùng Kim Phương, Nguyễn Quang Thi, Nguyễn Hữu Hùng, Đặng Quang Hiệu
2020 Tạp chí Khoa học và Công nghệ - Đại học Thái Nguyên  
-High robustness and less vulnerability to track loss, fragmentation, with ability to distinguish multiple objects of the same class.  ...  The reason is they focused on particular quality metrics like MOTA (Multiple Object Tracking Accuracy), MOTP (Multiple Object Tracking Precision) [1] , that was tested on most popular object classes (  ... 
doi:10.34238/tnu-jst.3678 fatcat:4wlaflmss5hfjdxxzlfcmtji5i

Adaptive Learning Rate for Visual Tracking Using Correlation Filters

C.S. Asha, A.V. Narasimhadhan
2016 Procedia Computer Science  
This method uses integral channel features in correlation filter framework with adaptive learning rate to efficiently track the object.  ...  Existing correlation filters use fixed learning rate to update filter template in every frame.  ...  The object with highest matching score is considered as tracked target. A fast normalized cross correlation is used to match template with the object in every frame 2 .  ... 
doi:10.1016/j.procs.2016.06.023 fatcat:ce7mrmtvpzfbtokfhcxqdy3dp4
« Previous Showing results 1 — 15 out of 259,946 results