检索结果-内蒙古大学图书馆

MSSTResNet-TLD: A robust tracking method based on tracking-learning-detection framework by using multi-scale spatio-temporal residual network feature model

引用

NEUROCOMPUTING 2019年 362卷 175-194页

作者： Liu, Bing Liu, Qiao Zhang, Taiping Yang, Yong Chongqing Univ Sci & Technol Coll Math & Big Data Chongqing Peoples R China Chongqing Univ Coll Comp Sci Chongqing Peoples R China Harbin Inst Technol Shenzhen Grad Sch Sch Comp Sci & Technol Harbin Heilongjiang Peoples R China China Mobile IOT Co Ltd Open Platform Dept Shenzhen Guangdong Peoples R China

The performance of tracking task is directly dependent on the appearance features of target object, a robust approach for constructing appearance features is crucial for adaptation the appearance change. To construct an accurate and robust appearance model for visual object tracking, we modify original deep residual learning network architecture and name it Multi-Scale Residual Network (MSResNet). The first video frame image and its related information of the current input video sequence are used to learn a multi-scale appearance model of target object and a loss function is minimized over the appearance features. Meanwhile, spatial information of each video frame and temporal information between successive video frames effectively combine with MSResNet. And thus the features are generated by MultiScale Spatio-Temporal Residual Network, which is named MSSTResNet feature model, can adapt to scale variation, illumination variation, background clutters, severe deformation of the target object, and so on. We implement a robust tracking method based on tracking-learning-detection framework by using our proposed MSSTResNet feature model and name it MSSTResNet-TLD tracker. Unlike the previous tracking methods, the MSResNet architecture is not offline pre-trained on a large auxiliary datasets but is directly learned end-to-end with a multi-task loss by using the current input video sequence. Furthermore, the multi-task loss function utilizes the classification loss and regression loss that is more accurate for target localization. Our experimental results demonstrate that the proposed tracking method outperforms the current state-of-the-art tracking methods on Visual Object tracking Benchmark (VOT-2016), Object tracking Benchmark (OTB-2015), and Unmanned Aerial Vehicles (UAV20L) test datasets. Furthermore, our MSSTResNet-TLD tracker is faster than previous most trackers based on deep Convolutional Neural Network (ConvNet or CNN) and our tracker is extremely robust to the ti

关键词： tracking-learning-detection Spatio-temporal feature Multi-scale feature Residual network

来源：评论

学校读者我要写书评

暂无评论

Robust Pedestrian tracking using improved tracking-learning-detection algorithm 16

Robust Pedestrian Tracking using improved Tracking-Learning-...

引用

10th Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP)

作者： Verma, Ritika Sreedevi, Indu Delhi Technol Univ Delhi India

ISBN: (纸本)9781450347532

Manual analysis of pedestrians for surveillance of large crowds in real time applications is not practical. tracking-learning-detection suggested by Kalal , Mikolajczyk and Matas [1] is one of the most prominent automatic object tracking system. TLD can track single object and can handle occlusion and appearance change but it suffers from limitations .In this paper, tracking of multiple objects and estimation of their trajectory is suggested using improved TLD. Feature tracking is suggested in place of grid based tracking to solve the limitation of tracking during out of plane rotation .This also leads to optimization of algorithm. Proposed algorithm also achieves auto-initialization with detection of pedestrians in the first frame which makes it suitable for real time pedestrian tracking.

关键词： tracking-learning-detection Pedestrian tracking Visual tracking

来源：评论

学校读者我要写书评

暂无评论

A Real Time Dual-Camera Surveillance System Based On tracking-learning-detection Algorithm

A Real Time Dual-Camera Surveillance System Based On Trackin...

引用

第25届中国控制与决策会议

作者： Jilin Hu Shiqiang Hu Zhuojin Sun School of Aeronautics and Astronautics Shanghai Jiao Tong University

ISBN: (纸本)9781467355339

In this paper, a master-slave dual camera (MSDC) system is designed to monitor the scene and acquire detailed information of distant object at the same time. The selected target, in the wide field of view (FOV) of master camera, is detected and tracked by algorithms of codebook and tracking-learning-detection (TLD), respectively. According to the method of Look Up Table (LUT), the slave camera (pan-tilt-zoom camera, PTZ camera) is directed to gaze at the target. In this way, the high resolution image of the object is available in the PTZ camera. Experiments of indoor and outdoor scenes have been conducted and the application of this dual-camera surveillance system in real time is illustrated. Furthermore, by comparing these two algorithms, it is found that TLD is more robust to illumination variation and the scene of multi motion objects. Besides, this set of system has the potential of being more intelligent through the employment of a video containing certain objects to initialize TLD, thus enabling the track of objects in real scene. Certain attempt in this point has been made in our experiment, which proves successful.

关键词： Long term tracking Dual-camera surveillance tracking-learning-detection codebook.

来源：评论

学校读者我要写书评

暂无评论

Improved tracking-learning-detection Human tracking Algorithm

Improved Tracking-Learning-Detection Human Tracking Algorith...

引用

International Conference on Information Technology and Management Innovation (ICITMI2012)

作者： Liang, Yan Pan, Feng Yang, Bo Li, Weixing Beijing Inst Technol Sch Automat 5 Zhongguancun South St Beijing 100081 Peoples R China 629 Xijin Rd Lanzhou Gansu Peoples R China

ISBN: (纸本)9783037855744

For the problem that appearance of human will be changed in motion, it is a challenge to those capabilities of tracker, related to adaptation of human appearance. In the paper, an improved tracking-learning-detection algorithm, including short-term-tracking, real-time-detection and online-learning, is proposed. Furthermore, scale-invariant feature is applied to the short-term-tracking to determine the location of target in consecutive frames. The video stream of the PETS2006 database is utilized to evaluate the proposed algorithm. Experiments results demonstrate the effectiveness, especially on the short-term-tracking performance.

关键词： Human tracking tracking-learning-detection Scale-invariant feature

来源：评论

学校读者我要写书评

暂无评论

A Robust Hand tracking Approach Based on Modified tracking-learning-detection Algorithm 8

引用

FTRA 8th International Conference on Multimedia and Ubiquitous Engineering (MUE)

作者： Shi, Haibin Lin, Zhanjian Tang, Weiwei Liao, Bruce Wang, Jolly Zheng, Lingxiang Xiamen Univ Sch Informat Sci & Engn Xiamen 361005 Peoples R China Intel China Co Ltd Shanghai 200241 Peoples R China

ISBN: (纸本)9783642549007;9783642548994

Hand tracking is an essential step for dynamic gesture recognition which catches a lot of attention in the field of gesture interaction. In this paper, we present a robust hand tracking approach for unconstrained videos based on modified tracking-learning-detection (TLD) algorithm, named BP-TLD. By introducing a skin color feature to the model, we make the algorithm more suitable for hand tracking. The experimental results show that BP-TLD has a better performance compared with other tracking algorithms such as TLD, MSEPF and Handvu. It indicates that our approach can meet the requirements of robustness and real-time better for the frontal-view vision-based human computer interactions.

关键词： Back Projection tracking-learning-detection Hand tracking

来源：评论

学校读者我要写书评

暂无评论

Online multi-object tracking combining optical flow and compressive tracking in Markov decision process

引用

JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION 2019年第Jan.期58卷 178-186页

作者： Yang, Tao Cappelle, Cindy Ruichek, Yassine El Bagdouri, Mohammed Univ Bourgogne Franche Comte Le2i EA7508 CNRS Arts & MetiersUTBM F-90010 Belfort France

Effective features are important for visual tracking, and efficiency also needs to be considered especially for multi-object tracking. Thanks to the simplicity, we think compressive sensing features are suitable for this task. In this paper, we use compressive sensing features to improve the Markov decision process (MDP) multi-object tracking framework. First, we design a single object tracker which uses the compressive tracking to correct the optical flow tracking and apply this tracker into the MDP tracking framework. The appearance model constructed during compressive tracking also helps for data association. In order to validate our method, we firstly test the designed single object tracker with a common dataset. Then, we test our multi-object tracking method for vehicle tracking. Finally, we analyze and test our approach in the multi-object tracking (MOT) benchmark for pedestrian tracking. The results show our approach performs superiorly against several state-of-the-art online multi-object trackers. (C) 2018 Elsevier Inc. All rights reserved.

关键词： Multi-object tracking Markov decision process tracking-learning-detection Compressive sensing features

来源：评论

学校读者我要写书评

暂无评论

An extended KCF tracking algorithm based on TLD structure in low frame rate videos

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2020年第29-30期79卷 20995-21012页

作者： Moridvaisi, Hooman Razzazi, Farbod Pourmina, Mohammad Ali Dousti, Massoud Islamic Azad Univ Dept Elect & Comp Engn Sci & Res Branch Tehran Iran

The KCF (Kernelized Correlation Filter) algorithm achieved a good performance on target tracking challenges. However, it still has some defects and problems of false tracking in low frame rate (LFR) scenarios, target scale variation, occlusion and out of view target, that exists in the correlation filter based methods. In this paper, we overcome the shortcomings of KCF tracking algorithm based on tracking-learning-detection (TLD) framework. The proposed algorithm trained two classifiers simultaneously, based on semi supervised co-training learning algorithm. Then, we comparatively evaluate the proposed method on TB-100 datasets by other trackers. The experimental results demonstrate that the precision and robustness of the improved tracking algorithm is higher than traditional KCF, TLD and the other top state-of-the-art tracking algorithms in LFR videos.

关键词： Low frame rate video processing Abrupt motion Large appearance changes tracking-learning-detection Kernelized correlation filter

来源：评论

学校读者我要写书评

暂无评论

Semi-supervised hyperspectral classification from a small number of training samples using a co-training approach

引用

ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING 2016年 121卷 60-76页

作者： Romaszewski, Michal Glomb, Przemyslaw Cholewa, Michal Polish Acad Sci Inst Theoret & Appl Informat Baltycka 5 PL-44100 Gliwice Poland

We present a novel semi-supervised algorithm for classification of hyperspectral data from remote sensors. Our method is inspired by the tracking-learning-detection (TLD) framework, originally applied for tracking objects in a video stream. TLD introduced the co-training approach called P-N learning, making use of two independent 'experts' (or learners) that scored samples in different feature spaces. In a similar fashion, we formulated the hyperspectral classification task as a co-training problem, that can be solved with the P-N learning scheme. Our method uses both spatial and spectral features of data, extending a small set of initial labelled samples during the process of region growing. We show that this approach is stable and achieves very good accuracy even for small training sets. We analyse the algorithm's performance on several publicly available hyperspectral data sets. (C) 2016 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved.

关键词： Hyperspectral classification Co-training tracking-learning-detection

来源：评论

学校读者我要写书评

暂无评论

Fish tracking Based on Improved TLD Algorithm in Real-World Underwater Environment

引用

MARINE TECHNOLOGY SOCIETY JOURNAL 2019年第3期53卷 80-89页

作者： Wang, Junhui Zhao, Meng Zou, Li Hu, Yi Cheng, Xuezhen Liu, Xiaofeng Shandong Univ Sci & Technol Coll Elect Engn & Automat 579 Qianwangang Rd Qingdao 266590 Shandong Peoples R China Hohai Univ Coll IoT Engn Nanjing Jiangsu Peoples R China

Fish tracking in the natural underwater environment is of great value for many applications, such as fish behavior analysis and the estimation of fish population density. Due to the variety of swimming postures, uneven illumination, and complicated background in the real-world underwater environment, most existing object tracking algorithms are not ideal for fish tracking. In this paper, a tracking algorithm based on TLD (tracking-learning-detection) is proposed, in which the shape and color features of fish are fully utilized to achieve accurate and rapid tracking. The proposed algorithm mainly focuses on the tracking module and detection module in the TLD framework. In the tracking module, a scale adaptive mean-shift algorithm is employed to deal with the variety of swimming postures and the uneven illumination by estimating the scale and using the color histogram of the background and target area. In the detection module, the Hu invariant moment of fish, which encodes the shape information, is used to detect the target quickly. The experiment results show that the proposed algorithm is effective and accurate for fish tracking in the real-world underwater environment.

关键词： real-world underwater environment fish tracking tracking-learning-detection scale adaptive mean shift Hu invariant

来源：评论

学校读者我要写书评

暂无评论

A Visual Object tracking Algorithm Based on Improved TLD

引用

ALGORITHMS 2020年第1期13卷 15页

作者： Zhen, Xinxin Fei, Shumin Wang, Yinmin Du, Wei Southeast Univ Sch Automat Nanjing 210096 Peoples R China Cent South Univ Sch Automat Changsha 410083 Peoples R China

Visual object tracking is an important research topic in the field of computer vision. tracking-learning-detection (TLD) decomposes the tracking problem into three modules-tracking, learning, and detection-which provides effective ideas for solving the tracking problem. In order to improve the tracking performance of the TLD tracker, three improvements are proposed in this paper. The built-in tracking module is replaced with a kernelized correlation filter (KCF) algorithm based on the histogram of oriented gradient (HOG) descriptor in the tracking module. Failure detection is added for the response of KCF to identify whether KCF loses the target. A more specific detection area of the detection module is obtained through the estimated location provided by the tracking module. With the above operations, the scanning area of object detection is reduced, and a full frame search is required in the detection module if objects fails to be tracked in the tracking module. Comparative experiments were conducted on the object tracking benchmark (OTB) and the results showed that the tracking speed and accuracy was improved. Further, the TLD tracker performed better in different challenging scenarios with the proposed method, such as motion blur, occlusion, and environmental changes. Moreover, the improved TLD achieved outstanding tracking performance compared with common tracking algorithms.

关键词： object tracking tracking-learning-detection kernelized correlation filters histogram of oriented gradient real time robustness

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：