東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Algorithms and Benchmarks for Robust...

Fan, Heng.

Linked to FindBook

Google Book

Amazon

博客來

Algorithms and Benchmarks for Robust Visual Object Tracking.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Algorithms and Benchmarks for Robust Visual Object Tracking./
Author:	Fan, Heng.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
Description:	174 p.
Notes:	Source: Dissertations Abstracts International, Volume: 83-02, Section: B.
Contained By:	Dissertations Abstracts International83-02B.
Subject:	Computer science. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28498258
ISBN:	9798522970987

Algorithms and Benchmarks for Robust Visual Object Tracking.
Fan, Heng.

Algorithms and Benchmarks for Robust Visual Object Tracking. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 174 p.

Source: Dissertations Abstracts International, Volume: 83-02, Section: B.

Thesis (Ph.D.)--State University of New York at Stony Brook, 2021.

This item must not be sold to any third party vendors.

Visual object tracking is an important problem in computer vision and has a long list of applications such as video surveillance, intelligent vehicles, human-machine interaction, etc. This thesis investigates tracking from two aspects: algorithms and benchmarks.In the algorithm part, I will first introduce a novel parallel tracking and verifying (PTAV) framework. The key idea is to decompose tracking into two components implemented on two separate threads. With a carefully designed collaboration mechanism, PTAV enjoys both the high efficiency provided by tracker and the strong discriminative power by verifier. Second, I investigate the recent Siamese tracking and advance it with a cascade framework. The resulted cascaded Siamese tracker improves discriminative ability by performing hard negative mining. Third, I explore temporal cues for improving tracking by designing a context-aware displacement attention (CADA) module to capture motion in videos. Finally, I explore the impact of the quality of proposal on tracking and design a cascaded regression-align classification module to improve proposal-based tracking.In the other part, I introduce three benchmarks, including LaSOT, TracKlinic and TOTB. LaSOT aims at providing a dedicated platform for training deep trackers. Meanwhile, it also serves as a testbed for evaluating and comparing different tracking algorithms. TracKlinic is utilized to diagnose a tracking algorithm under different challenge factors for improvement. Different from existing tracking benchmarks which focus on opaque object tracking, TOTB studies the problem of tracking transparent object by contributing the first transparent object tracking benchmark. In addition, a novel transparent object tracker is proposed to facilitate the development of algorithm on TOTB.

ISBN: 9798522970987Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Deep learning

Algorithms and Benchmarks for Robust Visual Object Tracking.
LDR:02972nmm a2200373 4500 001 2283495
005 20211029101501.5
008 220723s2021 ||||||||||||||||| ||eng d
020 $a 9798522970987
035 $a (MiAaPQ)AAI28498258
035 $a AAI28498258
040 $a MiAaPQ $c MiAaPQ
100 1 $a Fan, Heng. $3 3562461
245 1 0 $a Algorithms and Benchmarks for Robust Visual Object Tracking.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 174 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-02, Section: B.
500 $a Advisor: Ling, Haibin.
502 $a Thesis (Ph.D.)--State University of New York at Stony Brook, 2021.
506 $a This item must not be sold to any third party vendors.
520 $a Visual object tracking is an important problem in computer vision and has a long list of applications such as video surveillance, intelligent vehicles, human-machine interaction, etc. This thesis investigates tracking from two aspects: algorithms and benchmarks.In the algorithm part, I will first introduce a novel parallel tracking and verifying (PTAV) framework. The key idea is to decompose tracking into two components implemented on two separate threads. With a carefully designed collaboration mechanism, PTAV enjoys both the high efficiency provided by tracker and the strong discriminative power by verifier. Second, I investigate the recent Siamese tracking and advance it with a cascade framework. The resulted cascaded Siamese tracker improves discriminative ability by performing hard negative mining. Third, I explore temporal cues for improving tracking by designing a context-aware displacement attention (CADA) module to capture motion in videos. Finally, I explore the impact of the quality of proposal on tracking and design a cascaded regression-align classification module to improve proposal-based tracking.In the other part, I introduce three benchmarks, including LaSOT, TracKlinic and TOTB. LaSOT aims at providing a dedicated platform for training deep trackers. Meanwhile, it also serves as a testbed for evaluating and comparing different tracking algorithms. TracKlinic is utilized to diagnose a tracking algorithm under different challenge factors for improvement. Different from existing tracking benchmarks which focus on opaque object tracking, TOTB studies the problem of tracking transparent object by contributing the first transparent object tracking benchmark. In addition, a novel transparent object tracker is proposed to facilitate the development of algorithm on TOTB.
590 $a School code: 0771.
650 4 $a Computer science. $3 523869
650 4 $a Information technology. $3 532993
650 4 $a Artificial intelligence. $3 516317
650 4 $a Information science. $3 554358
650 4 $a Accuracy. $3 3559958
650 4 $a Deep learning. $3 3554982
650 4 $a Experiments. $3 525909
650 4 $a Neural networks. $3 677449
650 4 $a Classification. $3 595585
650 4 $a Algorithms. $3 536374
650 4 $a Localization. $3 3560711
650 4 $a Ablation. $3 3562462
650 4 $a Efficiency. $3 753744
653 $a Deep learning
653 $a Tracking technology
653 $a Benchmark reviews
653 $a Novel parallel tracking
690 $a 0984
690 $a 0489
690 $a 0800
690 $a 0723
710 2 $a State University of New York at Stony Brook. $b Computer Science. $3 1674709
773 0 $t Dissertations Abstracts International $g 83-02B.
790 $a 0771
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28498258