Cited 0 time in webofscience Cited 1 time in scopus

Learning Pixel-Level Semantic Affinity with Image-Level Supervision for Weakly Supervised Semantic Segmentation

Title
Learning Pixel-Level Semantic Affinity with Image-Level Supervision for Weakly Supervised Semantic Segmentation
Authors
Ahn, JiwoonKwak, Suha
Issue Date
2018-06-20
Citation
31st Meeting of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018, 4981-4990
Type
Conference
ISBN
9781538664209
ISSN
1063-6919
Abstract
The deficiency of segmentation labels is one of the main obstacles to semantic segmentation in the wild. To alleviate this issue, we present a novel framework that generates segmentation labels of images given their image-level class labels. In this weakly supervised setting, trained models have been known to segment local discriminative parts rather than the entire object area. Our solution is to propagate such local responses to nearby areas which belong to the same semantic entity. To this end, we propose a Deep Neural Network (DNN) called AffinityNet that predicts semantic affinity between a pair of adjacent image coordinates. The semantic propagation is then realized by random walk with the affinities predicted by AffinityNet. More importantly, the supervision employed to train AffinityNet is given by the initial discriminative part segmentation, which is incomplete as a segmentation annotation but sufficient for learning semantic affinities within small image areas. Thus the entire framework relies only on image-level class labels and does not require any extra data or annotations. On the PASCAL VOC 2012 dataset, a DNN learned with segmentation labels generated by our method outperforms previous models trained with the same level of supervision, and is even as competitive as those relying on stronger supervision. © 2018 IEEE.
URI
http://hdl.handle.net/20.500.11750/9704
DOI
10.1109/CVPR.2018.00523
Publisher
IEEE Computer Society
Files:
There are no files associated with this item.
Collection:
ETC2. Conference Papers


qrcode mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE