相关参考文献
注意：仅列出部分参考文献，下载原文获取全部文献信息。
Article
Computer Science, Artificial Intelligence
Xiaosong Zhang et al.
Summary: This study proposes a learning-to-match (LTM) method to break the Intersection-over-Union (IoU) restriction in object detection, allowing for flexible matching between objects and anchors. Experimental results demonstrate that LTM consistently outperforms traditional methods in object detection with lower computational cost.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Di Yuan et al.
Summary: This paper proposes an adaptive spatial-temporal context-aware (ASTCA) model within the DCF-based tracking framework to improve tracking accuracy and reduce background interference in UAV-tracking scenarios. The ASTCA model learns spatial-temporal context weight and incorporates spatial context information, which outperforms state-of-the-art tracking methods on standard UAV datasets.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2022)
Article
Automation & Control Systems
Bo Huang et al.
Summary: This article proposes a novel attentional transfer learning-based Siamese network (SiamATL) to inspire current tracker learning by fully exploiting previous knowledge. The introduction of an attentional online update strategy and instance-transfer discriminative correlation filter (ITDCF) enhances the tracker's distinguishing ability, ultimately achieving online tracking through a mutual compensation mechanism of cross-correlation matching and ITDCF detection in the decision-making subnetwork. Experiments demonstrate the superior performance of the approach compared to state-of-the-art tracking algorithms on multiple large-scale tracking datasets.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Yutao Cui et al.
Summary: This article introduces a fully convolutional online tracking framework (FCOT) that enables online learning for both classification and regression branches. The FCOT tackles the challenges of adapting to the regression branch by introducing an online regression model generator (RMG) and overcoming issues of target variation and error accumulation during tracking.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Christoph Mayer et al.
Summary: This paper proposes a tracker architecture based on a Transformer-based model prediction module, which captures global relations and learns more powerful target model predictions. Experimental results on multiple tracking datasets demonstrate the superior performance of the proposed tracker.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Computer Science, Information Systems
Kai Yang et al.
Summary: The Siamese network based on the region proposal network (RPN) has gained attention in visual tracking due to its accuracy and efficiency. However, the design of the RPN requires the selection of anchor boxes, which can be complex and affect the model's applicability. To address this, the proposed Siamese corner networks (SiamCorners) is an anchor-free tracker that uses a modified corner pooling layer to convert bounding box estimates into corner predictions. This approach makes the tracking algorithm more flexible and simpler while achieving comparable results to state-of-the-art trackers.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Article
Computer Science, Artificial Intelligence
Kai Yang et al.
Summary: Visual attention is widely utilized in visual tasks, with attention on PrDiMP and SiamBAN trackers. However, models like PrDiMP and SiamBAN do not update the target template online, and feature vectors are independent in IoU-Net and Siamese frameworks. The proposed template guidance attention network enhances robustness under challenging conditions and achieves state-of-the-art results on multiple benchmarks.
Article
Computer Science, Artificial Intelligence
Lianghua Huang et al.
Summary: GOT-10k is a large tracking database that covers over 560 classes of moving objects and 87 motion patterns, providing a unified training and evaluation platform for tracker development. Additionally, the introduction of a one-shot protocol for tracker evaluation avoids biased results towards familiar objects and promotes generalization in tracker development.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Zikun Zhou et al.
Summary: An adaptive ensemble perception tracking framework is proposed in this paper to address limitations in traditional tracking models. Through per-pixel prediction and confidence-guided ensemble prediction mechanisms, the algorithm demonstrates superior performance in accuracy and speed compared to state-of-the-art methods.
Proceedings Paper
Computer Science, Artificial Intelligence
Christoph Mayer et al.
Summary: Tracking objects with similar appearances to the target is a fundamental challenge in visual tracking. While many methods try to suppress distractors through stronger appearance models, this approach focuses on tracking the distractor objects in order to continue tracking the target effectively by using a learned association network and training strategy, achieving outstanding performance in experiments.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Zikun Zhou et al.
Summary: This paper proposes a method to track the salient local parts of the target to address targets with various appearance variations; through fine-grained saliency mining and saliency-association modeling, effective representations are learned, resulting in favorable tracking performance against state-of-the-art trackers.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Bo Huang et al.
Summary: SiamSTA is a Siamese network that uses spatio-temporal attention to robustly track UAVs by combining reliable local tracking and wide-range re-detection alternately. The integration of local tracking and global re-detection in SiamSTA demonstrates superior performance in anti-UAV scenarios, showcasing its strength in adapting to different tracking challenges.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Siyuan Cheng et al.
Summary: A novel Siamese relation network is proposed with efficient modules, Relation Detector and Refinement Module, to filter distractors from the background and generate accurate tracking results. A contrastive training strategy is introduced to improve the tracker's discriminability and robustness, leading to state-of-the-art results in various challenging scenarios.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Article
Computer Science, Artificial Intelligence
Di Yuan et al.
Summary: This paper proposes a self-supervised deep correlation tracker (self-SDCT) which achieves competitive tracking performance by utilizing a multi-cycle consistency loss and a similarity dropout strategy to optimize the learning of the feature extraction network.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Hamed Kiani Galoogahi et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)
Article
Computer Science, Artificial Intelligence
Yi Wu et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2015)
