Feature Rescaling and Fusion for Tiny Object Detection

doi:10.1109/ACCESS.2021.3074790

	Feature Rescaling and Fusion for Tiny Object Detection
	Liu, Jingwei 1,2; Gu, Yi 3; Han, Shumin 3; Zhang, Zhibin 1; Guo, Jiafeng 1; Cheng, Xueqi 1
	2021
发表期刊	IEEE ACCESS
ISSN	2169-3536
卷号	9 页码:62946-62955
摘要	Recent years have witnessed rapid developments on computer vision, however, there are still challenges in detecting tiny objects in a large-scale background. The tiny objects knowledge become sparse and weak due to their tiny size, which makes the tiny objects difficult to be detected with the common approaches. In this paper, a new network named Specific Characteristics based Feature Rescaling and Fusion (SFRF) is designed to detect tiny persons in a broad horizon and massive background. Different from the methods in general, a Nonparametric Adaptive Dense Perceiving Algorithm (NADPA) is designed to automatically select and generate a new resized feature map with the high density distribution of tiny objects. Then, a method called Many-For-One strategy is used for feature fusion of the feature pyramid network (FPN) layers to improve the feature representation and detection. Finally, an ensemble model method named hierarchical Coarse-to-fine mechanism is designed based on the proposed methods to further improve the performance. The experiments demonstrate that the proposed approach achieves an obvious performance improvement on tiny object detection than the existing approaches, and our approach has been awarded as the 1st-place in the first large-scale Tiny Object Detection (TOD) challenge.
关键词	Feature extraction Object detection Semantics Task analysis Training Spatial resolution Shape Tiny object detection nonparametric adaptive selection feature fusion feature pyramid network ensemble model
DOI	10.1109/ACCESS.2021.3074790
收录类别	SCI
语种	英语
WOS研究方向	Computer Science ; Engineering ; Telecommunications
WOS类目	Computer Science, Information Systems ; Engineering, Electrical & Electronic ; Telecommunications
WOS记录号	WOS:000645861200001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：5[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/16676
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Han, Shumin; Cheng, Xueqi
作者单位	1.Chinese Acad Sci, Inst Comp Technol, CAS Key Lab Network Data Sci & Technol, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China 3.Baidu Online Network Technol Beijing Co Ltd, Beijing 100085, Peoples R China
推荐引用方式 GB/T 7714	Liu, Jingwei,Gu, Yi,Han, Shumin,et al. Feature Rescaling and Fusion for Tiny Object Detection[J]. IEEE ACCESS,2021,9:62946-62955.
APA	Liu, Jingwei,Gu, Yi,Han, Shumin,Zhang, Zhibin,Guo, Jiafeng,&Cheng, Xueqi.(2021).Feature Rescaling and Fusion for Tiny Object Detection.IEEE ACCESS,9,62946-62955.
MLA	Liu, Jingwei,et al."Feature Rescaling and Fusion for Tiny Object Detection".IEEE ACCESS 9(2021):62946-62955.