Progressive Feature Learning for Facade Parsing With Occlusions

doi:10.1109/TIP.2022.3152004

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 三维可视计算

	Progressive Feature Learning for Facade Parsing With Occlusions
	Ma, Wenguang 1; Xu, Shibiao 2; Ma, Wei 1; Zhang, Xiaopeng3 ; Zha, Hongbin 4
发表期刊	IEEE TRANSACTIONS ON IMAGE PROCESSING
ISSN	1057-7149
	2022
卷号	31 页码:2081-2093
通讯作者	Ma, Wei(mawei@bjut.edu.cn)
摘要	Existing deep models for facade parsing often fail in classifying pixels in heavily occluded regions of facade images due to the difficulty in feature representation of these pixels. In this paper, we solve facade parsing with occlusions by progressive feature learning. To this end, we locate the regions contaminated by occlusions via Bayesian uncertainty evaluation on categorizing each pixel in these regions. Then, guided by the uncertainty, we propose an occlusion-immune facade parsing architecture in which we progressively re-express the features of pixels in each contaminated region from easy to hard. Specifically, the outside pixels, which have reliable context from visible areas, are re-expressed at early stages; the inner pixels are processed at late stages when their surroundings have been decontaminated at the earlier stages. In addition, at each stage, instead of using regular square convolution kernels, we design a context enhancement module (CEM) with directional strip kernels, which can aggregate structural context to re-express facade pixels. Extensive experiments on popular facade datasets demonstrate that the proposed method achieves state-of-the-art performance.
关键词	Uncertainty Bayes methods Convolutional neural networks Buildings Training Context modeling Representation learning Facade parsing occlusion feature representation manmade structure
DOI	10.1109/TIP.2022.3152004
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[62176010] ; National Natural Science Foundation of China[61771026] ; National Natural Science Foundation of China[U21A20515] ; National Natural Science Foundation of China[61971418] ; National Natural Science Foundation of China[61671451]
项目资助者	National Natural Science Foundation of China
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:000766266400001
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：6[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/47946
专题	多模态人工智能系统全国重点实验室_三维可视计算
通讯作者	Ma, Wei
作者单位	1.Beijing Univ Technol, Fac Informat Technol, Beijing 100124, Peoples R China 2.Beijing Univ Posts & Telecommun, Sch Artificial Intelligence, Beijing 100876, Peoples R China 3.Chinese Acad Sci, Inst Automat, Natl Lab Pattern Recognit, Beijing 100190, Peoples R China 4.Peking Univ, Sch Elect Engn & Comp Sci, Key Lab Machine Percept MOE, Beijing 100871, Peoples R China
推荐引用方式 GB/T 7714	Ma, Wenguang,Xu, Shibiao,Ma, Wei,et al. Progressive Feature Learning for Facade Parsing With Occlusions[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2022,31:2081-2093.
APA	Ma, Wenguang,Xu, Shibiao,Ma, Wei,Zhang, Xiaopeng,&Zha, Hongbin.(2022).Progressive Feature Learning for Facade Parsing With Occlusions.IEEE TRANSACTIONS ON IMAGE PROCESSING,31,2081-2093.
MLA	Ma, Wenguang,et al."Progressive Feature Learning for Facade Parsing With Occlusions".IEEE TRANSACTIONS ON IMAGE PROCESSING 31(2022):2081-2093.