Leveraging modality‐specific and shared features for RGB‐T salient object detection

Abstract Most of the existing RGB‐T salient object detection methods are usually based on dual‐stream encoding single‐stream decoding network architecture. These models always rely on the quality of fusion features, which often focus on modality‐shared features and overlook modality‐specific feature...

Full description

Saved in:
Bibliographic Details
Main Authors: Shuo Wang, Gang Yang, Qiqi Xu, Xun Dai
Format: Article
Language:English
Published: Wiley 2024-12-01
Series:IET Computer Vision
Subjects:
Online Access:https://doi.org/10.1049/cvi2.12307
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850059651952934912
author Shuo Wang
Gang Yang
Qiqi Xu
Xun Dai
author_facet Shuo Wang
Gang Yang
Qiqi Xu
Xun Dai
author_sort Shuo Wang
collection DOAJ
description Abstract Most of the existing RGB‐T salient object detection methods are usually based on dual‐stream encoding single‐stream decoding network architecture. These models always rely on the quality of fusion features, which often focus on modality‐shared features and overlook modality‐specific features, thus failing to fully utilise the rich information contained in multi‐modality data. To this end, a modality separate tri‐stream net (MSTNet), which consists of a tri‐stream encoding (TSE) structure and a tri‐stream decoding (TSD) structure is proposed. The TSE explicitly separates and extracts the modality‐shared and modality‐specific features to improve the utilisation of multi‐modality data. In addition, based on the hybrid‐attention and cross‐attention mechanism, we design an enhanced complementary fusion module (ECF), which fully considers the complementarity between the features to be fused and realises high‐quality feature fusion. Furthermore, in TSD, the quality of uni‐modality features is ensured under the constraint of supervision. Finally, to make full use of the rich multi‐level and multi‐scale decoding features contained in TSD, the authors design the adaptive multi‐scale decoding module and the multi‐stream feature aggregation module to improve the decoding capability. Extensive experiments on three public datasets show that the MSTNet outperforms 14 state‐of‐the‐art methods, demonstrating that this method can extract and utilise the multi‐modality information more adequately and extract more complete and rich features, thus improving the model's performance. The code will be released at https://github.com/JOOOOKII/MSTNet.
format Article
id doaj-art-7ef0a104dfe24ae29944b911265a682c
institution DOAJ
issn 1751-9632
1751-9640
language English
publishDate 2024-12-01
publisher Wiley
record_format Article
series IET Computer Vision
spelling doaj-art-7ef0a104dfe24ae29944b911265a682c2025-08-20T02:50:49ZengWileyIET Computer Vision1751-96321751-96402024-12-011881285129910.1049/cvi2.12307Leveraging modality‐specific and shared features for RGB‐T salient object detectionShuo Wang0Gang Yang1Qiqi Xu2Xun Dai3Northeastern University Shenyang ChinaNortheastern University Shenyang ChinaNortheastern University Shenyang ChinaNortheastern University Shenyang ChinaAbstract Most of the existing RGB‐T salient object detection methods are usually based on dual‐stream encoding single‐stream decoding network architecture. These models always rely on the quality of fusion features, which often focus on modality‐shared features and overlook modality‐specific features, thus failing to fully utilise the rich information contained in multi‐modality data. To this end, a modality separate tri‐stream net (MSTNet), which consists of a tri‐stream encoding (TSE) structure and a tri‐stream decoding (TSD) structure is proposed. The TSE explicitly separates and extracts the modality‐shared and modality‐specific features to improve the utilisation of multi‐modality data. In addition, based on the hybrid‐attention and cross‐attention mechanism, we design an enhanced complementary fusion module (ECF), which fully considers the complementarity between the features to be fused and realises high‐quality feature fusion. Furthermore, in TSD, the quality of uni‐modality features is ensured under the constraint of supervision. Finally, to make full use of the rich multi‐level and multi‐scale decoding features contained in TSD, the authors design the adaptive multi‐scale decoding module and the multi‐stream feature aggregation module to improve the decoding capability. Extensive experiments on three public datasets show that the MSTNet outperforms 14 state‐of‐the‐art methods, demonstrating that this method can extract and utilise the multi‐modality information more adequately and extract more complete and rich features, thus improving the model's performance. The code will be released at https://github.com/JOOOOKII/MSTNet.https://doi.org/10.1049/cvi2.12307computer visionlearning (artificial intelligence)
spellingShingle Shuo Wang
Gang Yang
Qiqi Xu
Xun Dai
Leveraging modality‐specific and shared features for RGB‐T salient object detection
IET Computer Vision
computer vision
learning (artificial intelligence)
title Leveraging modality‐specific and shared features for RGB‐T salient object detection
title_full Leveraging modality‐specific and shared features for RGB‐T salient object detection
title_fullStr Leveraging modality‐specific and shared features for RGB‐T salient object detection
title_full_unstemmed Leveraging modality‐specific and shared features for RGB‐T salient object detection
title_short Leveraging modality‐specific and shared features for RGB‐T salient object detection
title_sort leveraging modality specific and shared features for rgb t salient object detection
topic computer vision
learning (artificial intelligence)
url https://doi.org/10.1049/cvi2.12307
work_keys_str_mv AT shuowang leveragingmodalityspecificandsharedfeaturesforrgbtsalientobjectdetection
AT gangyang leveragingmodalityspecificandsharedfeaturesforrgbtsalientobjectdetection
AT qiqixu leveragingmodalityspecificandsharedfeaturesforrgbtsalientobjectdetection
AT xundai leveragingmodalityspecificandsharedfeaturesforrgbtsalientobjectdetection