Two-Stream Refinement Network for RGB-D Saliency Detection

Liu, Di; Hu, Yaosi; Zhang, Kao; Chen, Zhenzhong

doi:10.1109/icip.2019.8803653

Cited by 9 publications

(9 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fan et al [18] designed a depth depurator unit to filter out some lowquality depth maps. Most other models [5,7,25,33,35,37] employ cross-modal fusion at multiple scales using different integration strategies.…”

Section: Rgb-d Saliency Detectionmentioning

confidence: 99%

“…It has been widely applied in various vision-related tasks, such as image understanding [75], video/semantic segmentation [55,58], action recognition [51], [55], and person re-identification [67]. Although significant progress has been made, it is still challenging to accurately locate salient objects in com- images are fed into two separate network streams and then the fused high-level features are fed into a decoder (e.g., [4,5,25,37]). (b) Depth features are integrated into the RGB network using a tailor-maid subnetwork (e.g., [6,66,72]).…”

Section: Introductionmentioning

confidence: 99%

“…1 (a)). Besides, other methods carry out cross-modal fusion at multiple scales [4,5,7,25,27,37]. As a result, the complex correlations can be effectively exploited from the two modalities.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Specificity-preserving RGB-D Saliency Detection

Zhou¹,

Fan²,

Chen³

et al. 2021

Preprint

View full text Add to dashboard Cite

RGB-D saliency detection has attracted increasing attention, due to its effectiveness and the fact that depth cues can now be conveniently captured. Existing works often focus on learning a shared representation through various fusion strategies, with few methods explicitly considering how to preserve modality-specific characteristics. In this paper, taking a new perspective, we propose a specificitypreserving network (SP-Net) for RGB-D saliency detection, which benefits saliency detection performance by exploring both the shared information and modality-specific properties (e.g., specificity). Specifically, two modality-specific networks and a shared learning network are adopted to generate individual and shared saliency maps. A crossenhanced integration module (CIM) is proposed to fuse cross-modal features in the shared learning network, which are then propagated to the next layer for integrating crosslevel information. Besides, we propose a multi-modal feature aggregation (MFA) module to integrate the modalityspecific features from each individual decoder into the shared decoder, which can provide rich complementary multi-modal information to boost the saliency detection performance. Further, a skip connection is used to combine hierarchical features between the encoder and decoder layers. Experiments on six benchmark datasets demonstrate that our SP-Net outperforms other state-of-the-art methods.Code is available at: https://github.com/taozh2017/SPNet.

show abstract

Section: Rgb-d Saliency Detectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Specificity-preserving RGB-D Saliency Detection

Zhou¹,

Fan²,

Chen³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In [147], Wang et al proposed a twostream CNN by utilizing a fusion strategy. Similarly, in [148], Liu et al proposed a fusion-based two-stream network for RGBD saliency detection. The depth structure information help in the foreground and background identification.…”

Section: Rgbd Saliency Detectionmentioning

confidence: 99%

A brief survey of visual saliency detection

Ullah

Jian

Hussain

et al. 2020

Multimed Tools Appl

View full text Add to dashboard Cite

Salient object detection models mimic the behavior of human beings and capture the most salient region/object from the images or scenes. This field has many important applications in both computer vision and pattern recognition tasks. Despite hundreds of models proposed in this field, it still has a large room for research. This paper demonstrates a detailed overview of the recent progress of saliency detection models in terms of heuristic-based techniques and deep learning-based techniques. We have discussed and reviewed its co-related fields, such as Eyefixation-prediction, RGBD salient-object-detection, co-saliency object detection, and video-saliency-detection models. We have reviewed the key issues of the current saliency models and discussed future trends and recommendations. The broadly utilized datasets and assessment strategies are additionally investigated in this paper. PISA[71] CVPR 2013 SD+CP NA GR[72] SPL 2013 GC+CS M PCA[73] CVPR 2013 GC C++ COV[74] JOV 2013 LC+CS M RBD[18] CVPR 2014 BC M SLF[75] CVPR 2014 F+ B M PDE[68] CVPR 2014 CP+B+CLP NA ILP[67] ITOIP 2015 GC+SD NA

show abstract

“…Han et al [25] built a CNNs‐based RGB‐D saliency detection framework based on the cross‐view transfer and multi‐view fusion. Liu et al [26] designed a two‐stream refinement network for RGB‐D saliency detection. More recently, Chen et al used a cross‐modal distillation stream, the RGB and depth streams to design a three‐stream attention‐aware multi‐modal fusion network for salient object detection [27].…”

Section: Related Workmentioning

confidence: 99%