Aggregating multi-scale contextual features from multiple stages for semantic image segmentation

Semantic segmentation plays a vital role in image understanding. Recent studies have attempted to achieve precise pixel-level classification by using deep networks that provide hierarchical features. These methods are trying to effectively utilise multi-level features that are extracted from the dat...

Full description

Saved in:
Bibliographic Details
Main Authors: Dingchao Jiang, Hua Qu, Jihong Zhao, Jianlong Zhao, Meng-Yen Hsieh
Format: Article
Language:English
Published: Taylor & Francis Group 2021-07-01
Series:Connection Science
Subjects:
Online Access:http://dx.doi.org/10.1080/09540091.2020.1862059
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849701291860688896
author Dingchao Jiang
Hua Qu
Jihong Zhao
Jianlong Zhao
Meng-Yen Hsieh
author_facet Dingchao Jiang
Hua Qu
Jihong Zhao
Jianlong Zhao
Meng-Yen Hsieh
author_sort Dingchao Jiang
collection DOAJ
description Semantic segmentation plays a vital role in image understanding. Recent studies have attempted to achieve precise pixel-level classification by using deep networks that provide hierarchical features. These methods are trying to effectively utilise multi-level features that are extracted from the data and precisely reconstruct some characteristics of objects that are lost in producing high-level features. In this paper, we propose a multi-scale context U-net (MSCU-net) for semantic image segmentation. This network uses a multi-scale context block (MSCB) to aggregate multi-level features and employs the CRF layer to explicitly model the dependencies among pixels. This network significantly outperforms other state-of-the-art methods on both the PASCAL VOC 2012 and Cityscapes datasets.
format Article
id doaj-art-21c7d173f84141b884963f87b46e1c8c
institution DOAJ
issn 0954-0091
1360-0494
language English
publishDate 2021-07-01
publisher Taylor & Francis Group
record_format Article
series Connection Science
spelling doaj-art-21c7d173f84141b884963f87b46e1c8c2025-08-20T03:17:58ZengTaylor & Francis GroupConnection Science0954-00911360-04942021-07-0133360562210.1080/09540091.2020.18620591862059Aggregating multi-scale contextual features from multiple stages for semantic image segmentationDingchao Jiang0Hua Qu1Jihong Zhao2Jianlong Zhao3Meng-Yen Hsieh4Xi'an Jiaotong UniversityXi'an Jiaotong UniversityXi'an University of Posts and TelecommunicationsXi'an Jiaotong UniversityProvidence UniversitySemantic segmentation plays a vital role in image understanding. Recent studies have attempted to achieve precise pixel-level classification by using deep networks that provide hierarchical features. These methods are trying to effectively utilise multi-level features that are extracted from the data and precisely reconstruct some characteristics of objects that are lost in producing high-level features. In this paper, we propose a multi-scale context U-net (MSCU-net) for semantic image segmentation. This network uses a multi-scale context block (MSCB) to aggregate multi-level features and employs the CRF layer to explicitly model the dependencies among pixels. This network significantly outperforms other state-of-the-art methods on both the PASCAL VOC 2012 and Cityscapes datasets.http://dx.doi.org/10.1080/09540091.2020.1862059deep learningsemantic segmentationmulti-scale context
spellingShingle Dingchao Jiang
Hua Qu
Jihong Zhao
Jianlong Zhao
Meng-Yen Hsieh
Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
Connection Science
deep learning
semantic segmentation
multi-scale context
title Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
title_full Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
title_fullStr Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
title_full_unstemmed Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
title_short Aggregating multi-scale contextual features from multiple stages for semantic image segmentation
title_sort aggregating multi scale contextual features from multiple stages for semantic image segmentation
topic deep learning
semantic segmentation
multi-scale context
url http://dx.doi.org/10.1080/09540091.2020.1862059
work_keys_str_mv AT dingchaojiang aggregatingmultiscalecontextualfeaturesfrommultiplestagesforsemanticimagesegmentation
AT huaqu aggregatingmultiscalecontextualfeaturesfrommultiplestagesforsemanticimagesegmentation
AT jihongzhao aggregatingmultiscalecontextualfeaturesfrommultiplestagesforsemanticimagesegmentation
AT jianlongzhao aggregatingmultiscalecontextualfeaturesfrommultiplestagesforsemanticimagesegmentation
AT mengyenhsieh aggregatingmultiscalecontextualfeaturesfrommultiplestagesforsemanticimagesegmentation