Learning from Weak and Noisy Labels for Semantic Segmentation

A weakly supervised semantic segmentation (WSSS) method aims to learn a segmentation model from weak (image-level) as opposed to strong (pixel-level) labels. By avoiding the tedious pixel-level annotation process, it can exploit the unlimited supply of user-tagged images from media-sharing sites suc...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on pattern analysis and machine intelligence 2017-03, Vol.39 (3), p.486-500
Main Authors: Lu, Zhiwu, Fu, Zhenyong, Xiang, Tao, Han, Peng, Wang, Liwei, Gao, Xin
Format: Article
Language:English
Subjects:
Algorithms
Computational modeling
Image annotation
Image segmentation
label noise reduction
Labeling
Labels
Machine learning
Noise measurement
Noise reduction
Optimization
Pixels
Semantic segmentation
Semantics
sparse learning
Training
weakly supervised learning
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A weakly supervised semantic segmentation (WSSS) method aims to learn a segmentation model from weak (image-level) as opposed to strong (pixel-level) labels. By avoiding the tedious pixel-level annotation process, it can exploit the unlimited supply of user-tagged images from media-sharing sites such as Flickr for large scale applications. However, these `free' tags/labels are often noisy and few existing works address the problem of learning with both weak and noisy labels. In this work, we cast the WSSS problem into a label noise reduction problem. Specifically, after segmenting each image into a set of superpixels, the weak and potentially noisy image-level labels are propagated to the superpixel level resulting in highly noisy labels; the key to semantic segmentation is thus to identify and correct the superpixel noisy labels. To this end, a novel L 1 -optimisation based sparse learning model is formulated to directly and explicitly detect noisy labels. To solve the L 1 -optimisation problem, we further develop an efficient learning algorithm by introducing an intermediate labelling variable. Extensive experiments on three benchmark datasets show that our method yields state-of-the-art results given noise-free labels, whilst significantly outperforming the existing methods when the weak labels are also noisy.
ISSN:0162-8828
1939-3539
2160-9292
DOI:10.1109/TPAMI.2016.2552172