Semantic segmentation with context encoding and multi-path decoding
Semantic image segmentation aims to classify every pixel of a scene image to one of many classes. It implicitly involves object recognition, localization, and boundary delineation. In this paper, we propose a segmentation network called CGBNet to enhance the paring results by context encoding and mu...
| Main Authors: | , , , , |
|---|---|
| Other Authors: | |
| Format: | Journal Article |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/161039 |
| _version_ | 1811693275252260864 |
|---|---|
| author | Ding, Henghui Jiang, Xudong Shuai, Bing Liu, Ai Qun Wang, Gang |
| author2 | School of Electrical and Electronic Engineering |
| author_facet | School of Electrical and Electronic Engineering Ding, Henghui Jiang, Xudong Shuai, Bing Liu, Ai Qun Wang, Gang |
| author_sort | Ding, Henghui |
| collection | NTU |
| description | Semantic image segmentation aims to classify every pixel of a scene image to one of many classes. It implicitly involves object recognition, localization, and boundary delineation. In this paper, we propose a segmentation network called CGBNet to enhance the paring results by context encoding and multi-path decoding. We first propose a context encoding module that generates context contrasted local feature to make use of the informative context and the discriminative local information. This context encoding module greatly improves the segmentation performance, especially for inconspicuous objects. Furthermore, we propose a scale-selection scheme to selectively fuse the parsing results from different-scales of features at every spatial position. It adaptively selects appropriate score maps from rich scales of features. To improve the parsing results of boundary, we further propose a boundary delineation module that encourages the location-specific very-low-level feature near the boundaries to take part in the final prediction and suppresses them far from the boundaries. Without bells and whistles, the proposed segmentation network achieves very competitive performance in terms of all three different evaluation metrics consistently on the four popular scene segmentation datasets, Pascal Context, SUN-RGBD, Sift Flow, and COCO Stuff, ADE20K, and Cityscapes. |
| first_indexed | 2024-10-01T06:49:05Z |
| format | Journal Article |
| id | ntu-10356/161039 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T06:49:05Z |
| publishDate | 2022 |
| record_format | dspace |
| spelling | ntu-10356/1610392022-08-12T05:40:49Z Semantic segmentation with context encoding and multi-path decoding Ding, Henghui Jiang, Xudong Shuai, Bing Liu, Ai Qun Wang, Gang School of Electrical and Electronic Engineering Engineering::Electrical and electronic engineering Semantic Segmentation Context Encoding Semantic image segmentation aims to classify every pixel of a scene image to one of many classes. It implicitly involves object recognition, localization, and boundary delineation. In this paper, we propose a segmentation network called CGBNet to enhance the paring results by context encoding and multi-path decoding. We first propose a context encoding module that generates context contrasted local feature to make use of the informative context and the discriminative local information. This context encoding module greatly improves the segmentation performance, especially for inconspicuous objects. Furthermore, we propose a scale-selection scheme to selectively fuse the parsing results from different-scales of features at every spatial position. It adaptively selects appropriate score maps from rich scales of features. To improve the parsing results of boundary, we further propose a boundary delineation module that encourages the location-specific very-low-level feature near the boundaries to take part in the final prediction and suppresses them far from the boundaries. Without bells and whistles, the proposed segmentation network achieves very competitive performance in terms of all three different evaluation metrics consistently on the four popular scene segmentation datasets, Pascal Context, SUN-RGBD, Sift Flow, and COCO Stuff, ADE20K, and Cityscapes. Ministry of Education (MOE) This work was jointly supported by Singapore Ministry of Education Academic Research Fund (AcRF) Tier 3 Grant no: MOE2017-T3-1-001, and Zhejiang Leading Innovation Research Program 2018R01017. 2022-08-12T05:40:49Z 2022-08-12T05:40:49Z 2020 Journal Article Ding, H., Jiang, X., Shuai, B., Liu, A. Q. & Wang, G. (2020). Semantic segmentation with context encoding and multi-path decoding. IEEE Transactions On Image Processing, 29, 3520-3533. https://dx.doi.org/10.1109/TIP.2019.2962685 1057-7149 https://hdl.handle.net/10356/161039 10.1109/TIP.2019.2962685 31940532 2-s2.0-85079651750 29 3520 3533 en MOE2017-T3-1-001 IEEE Transactions on Image Processing © 2020 IEEE. All rights reserved. |
| spellingShingle | Engineering::Electrical and electronic engineering Semantic Segmentation Context Encoding Ding, Henghui Jiang, Xudong Shuai, Bing Liu, Ai Qun Wang, Gang Semantic segmentation with context encoding and multi-path decoding |
| title | Semantic segmentation with context encoding and multi-path decoding |
| title_full | Semantic segmentation with context encoding and multi-path decoding |
| title_fullStr | Semantic segmentation with context encoding and multi-path decoding |
| title_full_unstemmed | Semantic segmentation with context encoding and multi-path decoding |
| title_short | Semantic segmentation with context encoding and multi-path decoding |
| title_sort | semantic segmentation with context encoding and multi path decoding |
| topic | Engineering::Electrical and electronic engineering Semantic Segmentation Context Encoding |
| url | https://hdl.handle.net/10356/161039 |
| work_keys_str_mv | AT dinghenghui semanticsegmentationwithcontextencodingandmultipathdecoding AT jiangxudong semanticsegmentationwithcontextencodingandmultipathdecoding AT shuaibing semanticsegmentationwithcontextencodingandmultipathdecoding AT liuaiqun semanticsegmentationwithcontextencodingandmultipathdecoding AT wanggang semanticsegmentationwithcontextencodingandmultipathdecoding |