Edge-guided feature fusion network for RGB-T salient object detection
IntroductionRGB-T Salient Object Detection (SOD) aims to accurately segment salient regions in both visible light and thermal infrared images. However, many existing methods overlook the critical complementarity between these modalities, which can enhance detection accuracy.MethodsWe propose the Edg...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2024-12-01
|
| Series: | Frontiers in Neurorobotics |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnbot.2024.1489658/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850249723611447296 |
|---|---|
| author | Yuanlin Chen Zengbao Sun Cheng Yan Ming Zhao |
| author_facet | Yuanlin Chen Zengbao Sun Cheng Yan Ming Zhao |
| author_sort | Yuanlin Chen |
| collection | DOAJ |
| description | IntroductionRGB-T Salient Object Detection (SOD) aims to accurately segment salient regions in both visible light and thermal infrared images. However, many existing methods overlook the critical complementarity between these modalities, which can enhance detection accuracy.MethodsWe propose the Edge-Guided Feature Fusion Network (EGFF-Net), which consists of cross-modal feature extraction, edge-guided feature fusion, and salience map prediction. Firstly, the cross-modal feature extraction module captures and aggregates united and intersecting information in each local region of RGB and thermal images. Then, the edge-guided feature fusion module enhances the edge features of salient regions, considering that edge information is very helpful in refining significant area details. Moreover, a layer-by-layer decoding structure integrates multi-level features and generates the prediction of salience maps.ResultsWe conduct extensive experiments on three benchmark datasets and compare EGFF-Net with state-of-the-art methods. Our approach achieves superior performance, demonstrating the effectiveness of the proposed modules in improving both detection accuracy and boundary refinement.DiscussionThe results highlight the importance of integrating cross-modal information and edge-guided fusion in RGB-T SOD. Our method outperforms existing techniques and provides a robust framework for future developments in multi-modal saliency detection. |
| format | Article |
| id | doaj-art-33250b8fb8d24228bd1aaa78f382bb6c |
| institution | OA Journals |
| issn | 1662-5218 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neurorobotics |
| spelling | doaj-art-33250b8fb8d24228bd1aaa78f382bb6c2025-08-20T01:58:27ZengFrontiers Media S.A.Frontiers in Neurorobotics1662-52182024-12-011810.3389/fnbot.2024.14896581489658Edge-guided feature fusion network for RGB-T salient object detectionYuanlin ChenZengbao SunCheng YanMing ZhaoIntroductionRGB-T Salient Object Detection (SOD) aims to accurately segment salient regions in both visible light and thermal infrared images. However, many existing methods overlook the critical complementarity between these modalities, which can enhance detection accuracy.MethodsWe propose the Edge-Guided Feature Fusion Network (EGFF-Net), which consists of cross-modal feature extraction, edge-guided feature fusion, and salience map prediction. Firstly, the cross-modal feature extraction module captures and aggregates united and intersecting information in each local region of RGB and thermal images. Then, the edge-guided feature fusion module enhances the edge features of salient regions, considering that edge information is very helpful in refining significant area details. Moreover, a layer-by-layer decoding structure integrates multi-level features and generates the prediction of salience maps.ResultsWe conduct extensive experiments on three benchmark datasets and compare EGFF-Net with state-of-the-art methods. Our approach achieves superior performance, demonstrating the effectiveness of the proposed modules in improving both detection accuracy and boundary refinement.DiscussionThe results highlight the importance of integrating cross-modal information and edge-guided fusion in RGB-T SOD. Our method outperforms existing techniques and provides a robust framework for future developments in multi-modal saliency detection.https://www.frontiersin.org/articles/10.3389/fnbot.2024.1489658/fullsaliency detectionpixel featuresdynamic compensationedge informationfeature fusion |
| spellingShingle | Yuanlin Chen Zengbao Sun Cheng Yan Ming Zhao Edge-guided feature fusion network for RGB-T salient object detection Frontiers in Neurorobotics saliency detection pixel features dynamic compensation edge information feature fusion |
| title | Edge-guided feature fusion network for RGB-T salient object detection |
| title_full | Edge-guided feature fusion network for RGB-T salient object detection |
| title_fullStr | Edge-guided feature fusion network for RGB-T salient object detection |
| title_full_unstemmed | Edge-guided feature fusion network for RGB-T salient object detection |
| title_short | Edge-guided feature fusion network for RGB-T salient object detection |
| title_sort | edge guided feature fusion network for rgb t salient object detection |
| topic | saliency detection pixel features dynamic compensation edge information feature fusion |
| url | https://www.frontiersin.org/articles/10.3389/fnbot.2024.1489658/full |
| work_keys_str_mv | AT yuanlinchen edgeguidedfeaturefusionnetworkforrgbtsalientobjectdetection AT zengbaosun edgeguidedfeaturefusionnetworkforrgbtsalientobjectdetection AT chengyan edgeguidedfeaturefusionnetworkforrgbtsalientobjectdetection AT mingzhao edgeguidedfeaturefusionnetworkforrgbtsalientobjectdetection |