Article
Computer Science, Artificial Intelligence
Chang Xu, Qingwu Li, Qingkai Zhou, Xiongbiao Jiang, Dabing Yu, Yaqin Zhou
Summary: RGB-thermal salient object detection has unique advantages in handling challenging scenes, but existing methods often overlook the differences between imaging mechanisms and thermal image characteristics, resulting in unsatisfactory performance. To address this, an asymmetric cross-modal activation network is proposed to achieve more effective RGB-T SOD by exploiting the interactions of modality-specific features.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Guibiao Liao, Wei Gao, Ge Li, Junle Wang, Sam Kwong
Summary: This article proposes a novel CCFENet model for RGB-T salient object detection. The model addresses the issue of defective modalities using a cross-collaboration enhancement strategy (CCE) and aggregates multi-level complementary multi-modal features using a cross-scale cross-modal decoder (CCD). Experimental results demonstrate that CCFENet outperforms existing models on multiple datasets.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Nianchang Huang, Qiang Jiao, Qiang Zhang, Jungong Han
Summary: This study proposes a novel middle-level feature fusion structure for designing a lightweight RGB-D SOD model. The structure utilizes two shallow subnetworks to extract low- and middle-level unimodal features and fuses them once via a specially designed fusion module. Furthermore, a relation-aware multi-modal feature fusion module is introduced to capture cross-modal complementary information effectively. Experimental results demonstrate the effectiveness and superiority of the proposed method, which has significantly reduced parameters and runs at a high speed of 33 FPS.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Computer Science, Artificial Intelligence
Zun Li, Congyan Lang, Jun Hao Liew, Yidong Li, Qibin Hou, Jiashi Feng
Summary: The Feature Pyramid Network (FPN) based models have been effective in salient object detection, but often generate incomplete saliency maps due to indirect information propagation. The proposed Cross-layer Feature Pyramid Network (CFPN) improves progressive fusion in salient object detection by enabling direct cross-layer communication. Extensive experiments demonstrate that CFPN accurately locates salient regions and effectively segments object boundaries.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Zhou Huang, Huai-Xin Chen, Tao Zhou, Yun-Zhi Yang, Bi-Yuan Liu
Summary: Depth cues play a crucial role in salient object detection, but the quality of depth directly affects performance. Existing methods lack effective fusion modules to fully exploit the complex correlations between RGB images and depth cues. The proposed MCI-Net integrates high-level features and multi-level features to improve SOD performance.
Article
Computer Science, Software Engineering
Yue Gao, Meng Dai, Qing Zhang
Summary: This study proposes a novel RGB-D SOD method that addresses the challenges in existing methods through cross-modal feature interaction and multi-level feature fusion. Extensive experiments on benchmark datasets demonstrate the superiority of the proposed method over other state-of-the-art RGB-D SOD methods.
Article
Computer Science, Artificial Intelligence
Zhenyu Wang, Yunzhou Zhang, Yan Liu, Shichang Liu, Sonya Coleman, Dermot Kerr
Summary: This study introduces a novel salient object detection method based on a multi-feature fusion cross network, which addresses the issues of insufficient multi-level feature fusion ability and boundary blur, leading to improved detection performance.
IMAGE AND VISION COMPUTING
(2021)
Article
Computer Science, Information Systems
Shanqing Zhang, Yujie Chen, Yiheng Meng, Jianfeng Lu, Li Li, Rui Bai
Summary: This paper proposes a salient object detection algorithm that uses Gram matrix and its F norm to weigh the importance of each multi-level feature map, and uses weight to recursively fuse multi-level prediction results to generate the final saliency map. Experimental results demonstrate that the proposed method performs well in various scenes, especially in complex scenes.
MULTIMEDIA SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Yunqiu Lv, Bowen Liu, Jing Zhang, Yuchao Dai, Aixuan Li, Tong Zhang
Summary: The paper presents a novel semi-supervised active salient object detection method that optimizes two networks progressively and effectively improves labeling efficiency. Experimental results demonstrate that the method can achieve comparable detection results with fully-supervised deep models.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Artificial Intelligence
Yuzhu Ji, Haijun Zhang, Zequn Jie, Lin Ma, Q. M. Jonathan Wu
Summary: This article introduces a novel cross-attention based encoder-decoder model called CASNet for video salient object detection, incorporating self- and cross-attention modules to improve accuracy and consistency. Extensive experimental results demonstrate the effectiveness of CASNet model surpassing existing image- and video-based methods on multiple datasets.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Hongbo Bi, Ranwan Wu, Ziqi Liu, Huihui Zhu, Cong Zhang, Tian -Zhu Xiang
Summary: This paper proposes a cross-modal Hierarchical Interaction Network (HINet) to boost the salient object detection by excavating the cross-modal feature interaction and progressively multi-level feature fusion.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Jiang-Jiang Liu, Zhi-Ang Liu, Pai Peng, Ming-Ming Cheng
Summary: This paper introduces a new salient object detection method by optimizing the connections within the U-shape structure to achieve cross-scale information interaction and obtain more accurate features. A relative global calibration module is further designed to process multi-scale inputs simultaneously, improving feature aggregation and introducing only a few additional parameters.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Jin Zhang, Qiuwei Liang, Qianqian Guo, Jinyu Yang, Qing Zhang, Yanjiao Shi
Summary: The article introduces a Residual Refinement Network (R(2)Net) method for salient object detection, which improves the performance of salient object detection through the fusion strategy of multi-scale features and contextual features. Experimental results demonstrate that the proposed method performs excellently on multiple benchmark datasets.
IMAGE AND VISION COMPUTING
(2022)
Article
Computer Science, Information Systems
Nianchang Huang, Yi Liu, Qiang Zhang, Jungong Han
Summary: The study proposes a novel RGB-D salient object detection model that effectively combines cross-modal features from RGB-D images and unimodal features from RGB and depth images, achieving significant performance improvement on four benchmark datasets compared to state-of-the-art methods.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Engineering, Electrical & Electronic
Chenhao Zhang, Shanshan Gao, Deqian Mao, Yuanfeng Zhou
Summary: In the annotation procedure of salient object detection, we propose dynamic scale-aware learning and a dense sampling strategy, achieving targeted feature aggregation using a graph attention mechanism. Our method outperforms the current state-of-the-art approaches, as shown in extensive experiments on benchmark datasets.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)