Look, Perceive and Segment: Finding the Salient Objects in Images via Two-stream Fixation-Semantic CNNs

Chen, Xiaowu; Zheng, Anlin; Li, Jia; Lu, Feng

doi:10.1109/iccv.2017.119

Cited by 61 publications

(49 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We train our network on DUTS-TR [34] as used in [38,26,36]. For a more comprehensive demonstration, we also trained our network with VGG-16 [32] on MSRA10K [8] as used in [49,48,7,23] and on DUTS-TR as done in [50,26]. The training images are not done with any special treatment except the horizontal flipping.…”

Section: Methodsmentioning

confidence: 99%

“…We compare our approach denoted as BANet with 16 state-of-the-art methods, including KSR [37], HDHF [22], ELD [20], UCF [49], NLDF [29], Amulet [48], FSN [7], SRM [36], C2SNet [23], RA [6], Picanet [26], PAGRN [50], R3Net [9], DGRL [38], RFCN [35] and DSS [14]. For fair comparison, we obtain the saliency maps of these methods from authors or the deployment codes provided by authors.…”

Section: Comparisons With the State-of-the-artsmentioning

confidence: 99%

See 1 more Smart Citation

Selectivity or Invariance: Boundary-Aware Salient Object Detection

Zhang

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

181

141

View full text Add to dashboard Cite

Typically, a salient object detection (SOD) model faces opposite requirements in processing object interiors and boundaries. The features of interiors should be invariant to strong appearance change so as to pop-out the salient object as a whole, while the features of boundaries should be selective to slight appearance change to distinguish salient objects and background. To address this selectivityinvariance dilemma, we propose a novel boundary-aware network with successive dilation for image-based SOD. In this network, the feature selectivity at boundaries is enhanced by incorporating a boundary localization stream, while the feature invariance at interiors is guaranteed with a complex interior perception stream. Moreover, a transition compensation stream is adopted to amend the probable failures in transitional regions between interiors and boundaries. In particular, an integrated successive dilation module is proposed to enhance the feature invariance at interiors and transitional regions. Extensive experiments on six datasets show that the proposed approach outperforms 16 state-of-the-art methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Comparisons With the State-of-the-artsmentioning

confidence: 99%

Selectivity or Invariance: Boundary-Aware Salient Object Detection

Zhang

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

181

141

View full text Add to dashboard Cite

show abstract

“…Readers can refer to [6] for a comprehensive review on these methods. In recent years, CNNs have been successfully applied for saliency detection and have achieved substantial improvements due to their powerful representation ability [15]- [33], [40]- [42]. Many CNN-based works attempt to learn deep semantic properties of salient objects for further performance improvements.…”

Section: A Salient Object Detectionmentioning

confidence: 99%

“…Earlier methods [6]- [14] for salient object detection mostly employed primitive hand-crafted features; their performance is reasonable but far from satisfactory in complex scenes. Recently, deep convolutional neural networks (CNNs), thanks to their powerful feature representation abilities, have been successfully applied for salient object detection [15]- [33]. Yi The proposed multi-level contextual information integration jointly employs feature maps and side outputs.…”

Section: Introductionmentioning

confidence: 99%

Deep Salient Object Detection With Contextual Information Guidance

Liu

Han

Zhang

et al. 2020

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Integration of multi-level contextual information, such as feature maps and side outputs, is crucial for Convolutional Neural Networks (CNNs) based salient object detection. However, most existing methods either simply concatenate multilevel feature maps or calculate element-wise addition of multilevel side outputs, thus failing to take full advantages of them. In this work, we propose a new strategy for guiding multilevel contextual information integration, where feature maps and side outputs across layers are fully engaged. Specifically, shallower-level feature maps are guided by the deeper-level side outputs to learn more accurate properties of the salient object. In turn, the deeper-level side outputs can be propagated to highresolution versions with spatial details complemented by means of shallower-level feature maps. Moreover, a group convolution module is proposed with the aim to achieve high-discriminative feature maps, in which the backbone feature maps are divided into a number of groups and then the convolution is applied to the channels of backbone feature maps within each group. Eventually, the group convolution module is incorporated in the guidance module to further promote the guidance role. Experiments on three public benchmark datasets verify the effectiveness and superiority of the proposed method over the state-of-the-art methods.

show abstract

“…A comprehensive review of the works in saliency detection can be found in [7–9]. It should be mentioned that since 2015, a few supervised or unsupervised salient object detection methods have been proposed by utilising deep convolutional neural networks [10–19], autoencoders [20], and recurrent neural networks [21–23]. In general, a training‐based method has the potential to provide accurate results.…”

Section: Introductionmentioning

confidence: 99%