Meta-Seg: A Generalized Meta-Learning Framework for Multi-Class Few-Shot Semantic Segmentation

被引:16
作者
Cao, Zhiying [1 ,2 ,4 ]
Zhang, Tengfei [1 ,2 ,4 ]
Diao, Wenhui [1 ,2 ]
Zhang, Yue [1 ,2 ]
Lyu, Xiaode [1 ,3 ]
Fu, Kun [1 ,2 ,4 ]
Sun, Xian [1 ,2 ]
机构
[1] Chinese Acad Sci, Aerosp Informat Res Inst, Beijing 100194, Peoples R China
[2] Chinese Acad Sci, Inst Elect, Key Lab Network Informat Syst Technol NIST, Beijing 100190, Peoples R China
[3] Chinese Acad Sci, Inst Elect, Key Lab Microwave Imaging Technol, Beijing 100190, Peoples R China
[4] Univ Chinese Acad Sci, Sch Elect Elect & Commun Engn, Beijing 100190, Peoples R China
基金
中国国家自然科学基金;
关键词
Meta-learning; few-shot; semantic segmentation; NETWORKS;
D O I
10.1109/ACCESS.2019.2953465
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Semantic segmentation performs pixel-wise classification for given images, which can be widely used in autonomous driving, robotics, medical diagnostics and etc. The recent advanced approaches have witnessed rapid progress in semantic segmentation. However, these supervised learning based methods rely heavily on large-scale datasets to acquire strong generalizing ability, such that they are coupled with some constraints. Firstly, human annotation of pixel-level segmentation masks is laborious and time-consuming, which causes relatively expensive training data and make it hard to deal with urgent tasks in dynamic environment. Secondly, the outstanding performance of the above data-hungry methods will decrease with few available training examples. In order to overcome the limitations of the supervised learning semantic segmentation methods, this paper proposes a generalized meta-learning framework, named Meta-Seg. It consists of a meta-learner and a base-learner. Specifically, the meta-learner learns a good initialization and a parameter update strategy from a distribution of few-shot semantic segmentation tasks. The base-learner can be any semantic segmentation models theoretically and can implement fast adaptation (that is updating parameters with few iterations) under the guidance of the meta-learner. In this work, the successful semantic segmentation model FCN8s is integrated into Meta-Seg. Experiments on the famous few-shot semantic segmentation dataset PASCAL5(i) prove Meta-Seg is a promising framework for few-shot semantic segmentation. Besides, this method can provide with reference for the relevant researches of meta-learning semantic segmentation.
引用
收藏
页码:166109 / 166121
页数:13
相关论文
共 57 条
[1]  
[Anonymous], IEEE I CONF COMP VIS, DOI DOI 10.1109/ICCV.2017.324
[2]  
[Anonymous], PROC CVPR IEEE
[3]  
[Anonymous], 2018, ARXIV180705076
[4]  
[Anonymous], 2017, P IEEE C COMP VIS PA
[5]  
[Anonymous], 2015, PROC CVPR IEEE
[6]  
[Anonymous], PROC CVPR IEEE
[7]  
[Anonymous], 2017, P BRIT MACH VIS C
[8]  
[Anonymous], ADV NEURAL INFORM PR, DOI DOI 10.1109/TPAMI.2016.2577031
[9]  
[Anonymous], ADV NEURAL INFORM PR
[10]  
[Anonymous], 2017, P IEEE C COMP VIS PA