Decoupled Metric Network for Single-Stage Few-Shot Object Detection

doi:10.1109/TCYB.2022.3149825

CASIA OpenIR

	Decoupled Metric Network for Single-Stage Few-Shot Object Detection
	Lu, Yue 1,2; Chen, Xingyu 3; Wu, Zhengxing 1,2; Yu, Junzhi 1,4
发表期刊	IEEE TRANSACTIONS ON CYBERNETICS
ISSN	2168-2267
	2022-02-24
页码	12
通讯作者	Yu, Junzhi(junzhi.yu@ia.ac.cn)
摘要	Within the last few years, great efforts have been made to study few-shot learning. Although general object detection is advancing at a rapid pace, few-shot detection remains a very challenging problem. In this work, we propose a novel decoupled metric network (DMNet) for single-stage few-shot object detection. We design a decoupled representation transformation (DRT) and an image-level distance metric learning (IDML) to solve the few-shot detection problem. The DRT can eliminate the adverse effect of handcrafted prior knowledge by predicting objectness and anchor shape. Meanwhile, to alleviate the problem of representation disagreement between classification and location (i.e., translational invariance versus translational variance), the DRT adopts a decoupled manner to generate adaptive representations so that the model is easier to learn from only a few training data. As for a few-shot classification in the detection task, we design an IDML tailored to enhance the generalization ability. This module can perform metric learning for the whole visual feature, so it can be more efficient than traditional DML due to the merit of parallel inference for multiobjects. Based on the DRT and IDML, our DMNet efficiently realizes a novel paradigm for few-shot detection, called single-stage metric detection. Experiments are conducted on the PASCAL VOC dataset and the MS COCO dataset. As a result, our method achieves state-of-the-art performance in few-shot object detection. The codes are available at https://github.com/yrqs/DMNet.
关键词	Object detection Feature extraction Training Head Task analysis Shape Measurement Computer vision deep learning few-shot learning object detection
DOI	10.1109/TCYB.2022.3149825
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2019YFB1310300] ; National Natural Science Foundation of China[62022090]
项目资助者	National Key Research and Development Program of China ; National Natural Science Foundation of China
WOS研究方向	Automation & Control Systems ; Computer Science
WOS类目	Automation & Control Systems ; Computer Science, Artificial Intelligence ; Computer Science, Cybernetics
WOS记录号	WOS:000764856200001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/47965
通讯作者	Yu, Junzhi
作者单位	1.Chinese Acad Sci, Inst Automat, State Key Lab Management & Control Complex Syst, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China 3.Kuaishou Technol, Ytech, Beijing 100085, Peoples R China 4.Peking Univ, Coll Engn, BIC ESAT, Dept Adv Mfg & Robot,State Key Lab Turbulence & C, Beijing 100871, Peoples R China
推荐引用方式 GB/T 7714	Lu, Yue,Chen, Xingyu,Wu, Zhengxing,et al. Decoupled Metric Network for Single-Stage Few-Shot Object Detection[J]. IEEE TRANSACTIONS ON CYBERNETICS,2022:12.
APA	Lu, Yue,Chen, Xingyu,Wu, Zhengxing,&Yu, Junzhi.(2022).Decoupled Metric Network for Single-Stage Few-Shot Object Detection.IEEE TRANSACTIONS ON CYBERNETICS,12.
MLA	Lu, Yue,et al."Decoupled Metric Network for Single-Stage Few-Shot Object Detection".IEEE TRANSACTIONS ON CYBERNETICS (2022):12.