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, Information Systems
En Yu et al.
Summary: Existing online MOT algorithms typically involve detection and re-identification (ReID) as two subtasks. In order to improve efficiency, these subtasks are commonly integrated into a unified framework. However, since detection and ReID require different features, this creates an optimization contradiction during training. To address this issue, we propose the Global Context Disentangling (GCD) module to separate the learned representation into detection-specific and ReID-specific embeddings. Additionally, we develop the Guided Transformer Encoder (GTE) module to capture global semantic relations more effectively, improving the overall performance of the MOT framework.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Artificial Intelligence
Yihong Xu et al.
Summary: Transformers have shown excellent performance in various tasks, but there is still room for improvement in multiple-object tracking. This paper proposes TransCenter, a transformer-based architecture that accurately tracks all objects using dense representations while maintaining reasonable runtime. TransCenter achieves remarkable performance improvements and outperforms state-of-the-art methods in two standard MOT benchmarks by utilizing dense image-related detection queries and efficient sparse tracking queries.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Yifu Zhang et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Fangao Zeng et al.
Summary: This paper proposes MOTR, an extension of DETR that models tracked instances in the entire video using track queries and incorporates techniques such as tracklet-aware label assignment, temporal aggregation network, and collective average loss to enhance temporal relation modeling. Experimental results demonstrate that MOTR achieves significant improvements in multiple-object tracking.
COMPUTER VISION - ECCV 2022, PT XXVII
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Xingyi Zhou et al.
Summary: This paper presents a novel transformer-based architecture for global multi-object tracking. The proposed architecture achieves global multi-object tracking without intermediate pairwise grouping or combinatorial association. Experimental results demonstrate that the architecture achieves competitive performance on the MOT17 benchmark and can seamlessly integrate into large-vocabulary detectors.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Tim Meinhardt et al.
Summary: This study proposes an end-to-end trainable multi-object tracking approach called TrackFormer, based on an encoder-decoder Transformer architecture. TrackFormer achieves outstanding performance in track initialization, identity, and spatio-temporal trajectory reasoning, and introduces the attention mechanism. Through self- and encoder-decoder attention on global frame-level features, additional graph optimization or modeling of motion and/or appearance is omitted.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Jiarui Cai et al.
Summary: This study proposes an online tracking algorithm that utilizes a large spatio-temporal memory to link objects after a long time span. The algorithm achieves excellent performance in object detection and data association tasks.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Computer Science, Artificial Intelligence
Jonathon Luiten et al.
Summary: The higher order tracking accuracy (HOTA) is a novel evaluation metric for multi-object tracking that balances accurate detection, association, and localization. It decomposes into sub-metrics to evaluate different error types separately, providing clear analysis of tracking performance. The HOTA scores align better with human visual evaluation of tracking performance compared to established metrics.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2021)
Article
Computer Science, Artificial Intelligence
Fan Yang et al.
Summary: The refinement method was studied for Multiple Object Tracking (MOT) tasks, defining Mix-up Error and Cut-off Error in imperfect tracklets, proposing the ReMOT framework to improve appearance features by splitting and merging tracklets, significantly improving MOT results, and assisting semi-automatic MOT data annotation.
IMAGE AND VISION COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Yifu Zhang et al.
Summary: Multi-object tracking is a crucial problem in computer vision, and formulating it as multi-task learning of object detection and re-ID in a single network can lead to joint optimization of the two tasks. However, competition between the tasks needs to be addressed, and the proposed FairMOT method based on CenterNet architecture achieves high accuracy for both detection and tracking through detailed designs and empirical studies.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Tarasha Khurana et al.
Summary: The research introduces a new approach to detecting invisible objects, focusing on the case of people. By treating occluded object detection as a short-term forecasting challenge and building dynamic models, the tracking and detection of occluded objects is achieved. The performance improvement is significant compared to existing technologies.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Pavel Tokmakov et al.
Summary: Traditional online multi-object tracking methods heavily rely on instantaneous observations, making them prone to failure when objects are not fully visible. In contrast, humans track objects based on the concept of object permanence. This study introduces an end-to-end trainable approach for joint object detection and tracking, which utilizes a large synthetic dataset for training and improves model robustness against occlusions.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Automation & Control Systems
Yongxin Wang et al.
Summary: Object detection and data association are critical components in multi-object tracking systems. Recent works have shown that simultaneously optimizing detection and data association modules under a joint MOT framework can lead to improved performance. This study proposes a new instance of joint MOT approach based on Graph Neural Networks, which can model relations between variable-sized objects in both spatial and temporal domains, leading to state-of-the-art performance for both detection and MOT tasks.
2021 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2021)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Linyu Zheng et al.
Summary: This paper proposes a novel and end-to-end trainable MOT architecture that extends CenterNet by adding an SOT branch for tracking objects in parallel. This allows the MOT task to benefit from the strong discriminative power of SOT methods in an effective and efficient way, achieving high performance with a frame rate of 16 FPS on MOT17.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Bing Shuai et al.
Summary: This paper introduces a region-based Siamese Multi-Object Tracking network called SiamMOT to improve online multi-object tracking by modeling motion. Experimental results show that SiamMOT performs superiorly in MOT and outperforms the winners of ACM MM'20 HiEve Grand Challenge on the HiEve dataset.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Article
Computer Science, Artificial Intelligence
Peng Tang et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2020)
Article
Computer Science, Information Systems
Yang Zhang et al.
IEEE INTERNET OF THINGS JOURNAL
(2020)
Proceedings Paper
Engineering, Electrical & Electronic
Peng Chu et al.
2019 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV)
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Shanshan Zhang et al.
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Kaiming He et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Fan Yang et al.
2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2016)
Article
Computer Science, Artificial Intelligence
Anton Milan et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2014)
Proceedings Paper
Computer Science, Artificial Intelligence
Seung-Hwan Bae et al.
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2014)
Proceedings Paper
Computer Science, Artificial Intelligence
Caglayan Dicle et al.
2013 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2013)
Article
Engineering, Electrical & Electronic
Keni Bernardin et al.
EURASIP JOURNAL ON IMAGE AND VIDEO PROCESSING
(2008)