Multilevel Features-Guided Network for Few-Shot Segmentation
The purpose of few-shot semantic segmentation is to segment unseen classes with only a few labeled samples. However, most methods ignore the guidance of low-level features for segmentation, leading to unsatisfactory results. Therefore, we propose a multilevel features-guided network using convolutio...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/11/19/3195 |
| _version_ | 1797479719477706752 |
|---|---|
| author | Chenjing Xin Xinfu Li Yunfeng Yuan |
| author_facet | Chenjing Xin Xinfu Li Yunfeng Yuan |
| author_sort | Chenjing Xin |
| collection | DOAJ |
| description | The purpose of few-shot semantic segmentation is to segment unseen classes with only a few labeled samples. However, most methods ignore the guidance of low-level features for segmentation, leading to unsatisfactory results. Therefore, we propose a multilevel features-guided network using convolutional neural network techniques, which fully utilizes features from each level. It includes two novel designs: (1) a similarity-guided feature reinforcement module (SRM), which uses features from different levels, it enables sufficient guidance from the support set to the query set, thus avoiding the situation that some feature information is ignored in deep network computation, (2) a method that bridges query features at each level to the decoder to guide the segmentation, making full use of local and edge information to improve model performance. We experiment on PASCAL-5<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> and COCO-20<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> datasets to demonstrate the effectiveness of the model, the results in 1-shot setting and 5-shot setting on PASCAL-5<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> are 64.7% and 68.0%, which are 3.9% and 6.1% higher than the baseline model, respectively, and the results on the COCO-20<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> are also improved. |
| first_indexed | 2024-03-09T21:50:52Z |
| format | Article |
| id | doaj.art-6295af36baa74bf486b981f6dff8575d |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-09T21:50:52Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-6295af36baa74bf486b981f6dff8575d2023-11-23T20:07:56ZengMDPI AGElectronics2079-92922022-10-011119319510.3390/electronics11193195Multilevel Features-Guided Network for Few-Shot SegmentationChenjing Xin0Xinfu Li1Yunfeng Yuan2School of Cyber Security and Computer, Hebei University, Baoding 071002, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding 071002, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding 071002, ChinaThe purpose of few-shot semantic segmentation is to segment unseen classes with only a few labeled samples. However, most methods ignore the guidance of low-level features for segmentation, leading to unsatisfactory results. Therefore, we propose a multilevel features-guided network using convolutional neural network techniques, which fully utilizes features from each level. It includes two novel designs: (1) a similarity-guided feature reinforcement module (SRM), which uses features from different levels, it enables sufficient guidance from the support set to the query set, thus avoiding the situation that some feature information is ignored in deep network computation, (2) a method that bridges query features at each level to the decoder to guide the segmentation, making full use of local and edge information to improve model performance. We experiment on PASCAL-5<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> and COCO-20<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> datasets to demonstrate the effectiveness of the model, the results in 1-shot setting and 5-shot setting on PASCAL-5<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> are 64.7% and 68.0%, which are 3.9% and 6.1% higher than the baseline model, respectively, and the results on the COCO-20<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mi>i</mi></msup></semantics></math></inline-formula> are also improved.https://www.mdpi.com/2079-9292/11/19/3195few-shot learningfew-shot segmentationscene understandingsemantic segmentation |
| spellingShingle | Chenjing Xin Xinfu Li Yunfeng Yuan Multilevel Features-Guided Network for Few-Shot Segmentation Electronics few-shot learning few-shot segmentation scene understanding semantic segmentation |
| title | Multilevel Features-Guided Network for Few-Shot Segmentation |
| title_full | Multilevel Features-Guided Network for Few-Shot Segmentation |
| title_fullStr | Multilevel Features-Guided Network for Few-Shot Segmentation |
| title_full_unstemmed | Multilevel Features-Guided Network for Few-Shot Segmentation |
| title_short | Multilevel Features-Guided Network for Few-Shot Segmentation |
| title_sort | multilevel features guided network for few shot segmentation |
| topic | few-shot learning few-shot segmentation scene understanding semantic segmentation |
| url | https://www.mdpi.com/2079-9292/11/19/3195 |
| work_keys_str_mv | AT chenjingxin multilevelfeaturesguidednetworkforfewshotsegmentation AT xinfuli multilevelfeaturesguidednetworkforfewshotsegmentation AT yunfengyuan multilevelfeaturesguidednetworkforfewshotsegmentation |