ADT: Object Tracking Algorithm Based on Adaptive Detection

被引:5
作者
Ming, Yue [1 ]
Zhang, Yashu [1 ]
机构
[1] Beijing Univ Posts & Telecommun, Sch Elect Engn, Beijing Key Lab Work Safety Intelligent Monitorin, Beijing 100876, Peoples R China
来源
IEEE ACCESS | 2020年 / 8卷 / 08期
基金
中央高校基本科研业务费专项资金资助; 北京市自然科学基金;
关键词
Recurrent neural network; adaptive detection; object tracking; correlation filtering; model compression; SIAMESE NETWORKS; SCALE;
D O I
10.1109/ACCESS.2020.2981525
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Object tracking is one of the most fundamental and important fields in computer vision with a wide range of applications. Although great progress has been made in object tracking combined with detection, there is still enormous challenges in real-time applications and for the computer cannot effectively capture the temporal correlations of targets and background clutter. In order to improve the performance of tracking algorithms under complex unconstrained conditions, we propose a novel tracking framework based on adaptive detection, called adaptive detection tracking (ADT). First, we exploit the temporal correlation of the recurrent neural network to predict the target & x2019;s motion direction and efficiently update the region of interest (RoI) in the narrow range of the next frame. Then, the algorithm utilizes the correlation filter to initialize the defined region of interest based on the threshold. If the Interaction of Union (IoU) of the predicted bounding box and the groundtruth bounding box is greater than the set threshold, the predicted bounding box will be directly output as the tracking results, whereas the detection is adaptively carried out in the determined RoI. Finally, the predicted bounding box refines the direction model as the input of the next frame to complete the whole tracking flow. Our proposed adaptive detection tracking mechanism can efficiently realize non-frame-by-frame adaptive detection with excellent tracking accuracy and is more robust in the unconstrained scenes, especially for occlusion. Comprehensive experiments demonstrate that our approach consistently achieves state-of-the-art results and runs in real-time on six large tracking benchmarks, including OTB100, VOT2016, VOT2017, TC128, UAV123 and LaSOT datasets.
引用
收藏
页码:56666 / 56679
页数:14
相关论文
共 75 条
  • [41] One-Shot Scale and Angle Estimation for Fast Visual Object Tracking
    Lee, Dong-Hyun
    [J]. IEEE ACCESS, 2019, 7 : 55477 - 55484
  • [42] On-Road Pedestrian Tracking Across Multiple Driving Recorders
    Lee, Kuan-Hui
    Hwang, Jenq-Neng
    [J]. IEEE TRANSACTIONS ON MULTIMEDIA, 2015, 17 (09) : 1429 - 1438
  • [43] SiamRPN plus plus : Evolution of Siamese Visual Tracking with Very Deep Networks
    Li, Bo
    Wu, Wei
    Wang, Qiang
    Zhang, Fangyi
    Xing, Junliang
    Yan, Junjie
    [J]. 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, : 4277 - 4286
  • [44] High Performance Visual Tracking with Siamese Region Proposal Network
    Li, Bo
    Yan, Junjie
    Wu, Wei
    Zhu, Zheng
    Hu, Xiaolin
    [J]. 2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2018, : 8971 - 8980
  • [45] Learning Spatial-Temporal Regularized Correlation Filters for Visual Tracking
    Li, Feng
    Tian, Cheng
    Zuo, Wangmeng
    Zhang, Lei
    Yang, Ming-Hsuan
    [J]. 2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2018, : 4904 - 4913
  • [46] GradNet: Gradient-Guided Network for Visual Object Tracking
    Li, Peixia
    Chen, Boyu
    Ouyan, Wanli
    Wang, Dong
    Yang, Xiaoyun
    Lu, Huchuan
    [J]. 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, : 6161 - 6170
  • [47] Deep visual tracking: Review and experimental comparison
    Li, Peixia
    Wang, Dong
    Wang, Lijun
    Lu, Huchuan
    [J]. PATTERN RECOGNITION, 2018, 76 : 323 - 338
  • [48] Target-Aware Deep Tracking
    Li, Xin
    Ma, Chao
    Wu, Baoyuan
    He, Zhenyu
    Yang, Ming-Hsuan
    [J]. 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, : 1369 - 1378
  • [49] Encoding Color Information for Visual Tracking: Algorithms and Benchmark
    Liang, Pengpeng
    Blasch, Erik
    Ling, Haibin
    [J]. IEEE TRANSACTIONS ON IMAGE PROCESSING, 2015, 24 (12) : 5630 - 5644
  • [50] Discriminative Correlation Filter with Channel and Spatial Reliability
    Lukezic, Alan
    Vojir, Tomas
    Zajc, Luka Cehovin
    Matas, Jiri
    Kristan, Matej
    [J]. 30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017), 2017, : 4847 - 4856