Discriminative object tracking with subspace representation

Devi, Rajkumari Bidyalakshmi; Chanu, Yambem Jina; Singh, Khumanthem Manglem

doi:10.1007/s00371-020-01862-0

Cited by 9 publications

(2 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At present, target tracking algorithms are mainly divided into two categories: discriminative trackers and generative trackers.The generative trackers establishes a target model through online learning, and then uses the model search to reestablish the image area with the smallest error to loacte the target. This type of method does not consider the background information of the target, and the image information is not well used, such as mean shift, Kalman filter algorithm.Discriminative trackers regards the target tracking as a binary classification problem, and extracts the target and background information to train the classifier by separating the target from the background of the image sequence [4].…”

Section: Target Tracking Methodsmentioning

confidence: 99%

Target localization network based on optical flow for object tracking

Chen

et al. 2023

Third International Conference on Artificial Intelligence and Computer Engineering (ICAICE 2022)

View full text Add to dashboard Cite

Object tracking is a very challenging branch in the field of computer vision and plays an extremely important role in different fields. However, most algorithms still have no good robustness when the target is out of view, fast motion and low resulution. Therefore, this paper proposes a method based on siamese architecture and optical flow to address these situations. The tracking framework is mainly divided into tracking network and target localization network based on optical flow. The tracking network uses SiamBAN and the target localization network uses LiteFlowNet3 to estimate the optical flow information of target. The proposed method in this paper expands the search area of SiamBAN, and uses the target localization network to reposition the target when tracking fails. The target localization network can be easily embed into the existing siamese network to provide more accurate location of target during the tracking process. Experiments results on VOT2016, VOT2018 show that this method has good robustness against extreme conditions.

show abstract

Section: Target Tracking Methodsmentioning

confidence: 99%

Target localization network based on optical flow for object tracking

Chen

et al. 2023

Third International Conference on Artificial Intelligence and Computer Engineering (ICAICE 2022)

View full text Add to dashboard Cite

show abstract

“…Visual image tracking algorithm models the appearance and motion information of the target by using the features and background information of the video or image sequence, to predict the target motion state and obtain the final position of the target. Visual image tracking algorithms can be divided into two categories from the perspective of model construction: generative algorithms [4] and discriminative algorithms [5]. The generative model algorithm does not consider the background information and achieves the purpose of tracking by matching the established model with the target category, such as the algorithm based on sparse representation and the algorithm based on subspace.…”

Section: Introductionmentioning

confidence: 99%

Research on Background Learning Correlation Filtering Algorithm With Multi-Feature Fusion

2023

View full text Add to dashboard Cite

Aiming at the problems of occlusion, drift, and background change in target tracking, a background learning correlation filtering algorithm based on multi-feature fusion is proposed. In the framework of correlation filtering, multi-feature fusion, multi-template update, and background learning regularization are used to improve the performance of the filter in the problem of template contamination and object occlusion. The fast directional gradient histogram (FHOG), color feature (CN), and texture feature (ULBP) were extracted, and the feature channels were connected in series. Then the depth features of Conv4-4 and Conv5-4 layers were extracted through the VGG-19 network, and the appearance model of the target was constructed. To reduce the sensitivity of the filter to the sudden change of background, a background learning filter is constructed, and the alternate direction multiplier method (ADMM) is used to speed up the calculation of the filter. In the model update stage, aiming at the problem of pollution of the original template caused by target occlusion, a high-confidence multi-template fusion update strategy is proposed by fusing the template with the highest confidence in the current frame, the previous frame, and the history frame. Finally, the proposed algorithm is tested on OTB50, OTB100, UAV123, and TC128 experimental data sets, and some classical and latest algorithms. The experimental results show that the tracking accuracy and robustness of the correlation filtering algorithm are improved.INDEX TERMS Target tracking, Machine vision, Correlation filtering, Background learning, More templates.

show abstract

Vehicle counting method based on attention mechanism SSD and state detection

Liang

Huang

2021

Vis Comput

View full text Add to dashboard Cite

Discriminative object tracking with subspace representation

Cited by 9 publications

References 34 publications

Target localization network based on optical flow for object tracking

Target localization network based on optical flow for object tracking

Research on Background Learning Correlation Filtering Algorithm With Multi-Feature Fusion

Vehicle counting method based on attention mechanism SSD and state detection

Contact Info

Product

Resources

About