.png){kind=icon}
返回
A UAV to UAV tracking benchmark
.png){kind=icon}
PRE
AI
DOI：10.1016/j.knosys.2022.110197.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Visual object tracking has achieved remarkable progress in recent years. However, current research in the vision community mainly focuses on tracking of generic objects, while less attention is paid to tracking unmanned aerial vehicle (UAV) especially from a moving platform. In this work, we explore this problem by constructing a UAV to UAV (UAV2UAV) tracking dataset. Specifically, this dataset consists of 44 videos (23k frames). Each video is manually annotated with bounding boxes. To the best of our knowledge, this dataset is the first work dedicated to UAV2UAV tracking. We extensively evaluate 19 state-of-the-art trackers using both hand-crafted feature based and deep-learning based approaches to understand how existing tracking methods perform and to provide a baseline for future works. The evaluation results show that more research works are needed to improve UAV2UAV tracking. In addition, we introduce a novel tracker, which leverages data augmentation technique for UAV tracking to encourage future research. By releasing this dataset, we expect to facilitate future studies and applications of UAV2UAV tracking in both academia and industry communities. Our dataset is available at: https://github.com/hapless19/UAV2UAV-dataset.(c) 2022 Elsevier B.V. All rights reserved.
Keyword:
UAV2UAV tracking
Data augmentation
Correlation filter tracking
期刊
K
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
Role of Urban Landscapes in Changing the Irrigation Water Requirements in Arid Climate.png)
Geosciences
OAAI
GeosciencesIF0
OAAI分享收藏Geriatric Assessments and Medication Therapy Management Approaches in Elderly Oncology Patients
PREAI
分享收藏
查看更多内容