Depth-Induced Multi-Scale Recurrent Attention Network for Saliency Detection

By Yongri Piao, Jingjing Li, Miao Zhang, Huchuan Lu. IEEE International Conference on Computer Vision, 2019

468 citations recorded by the source corpus.

In this work, we propose a novel depth-induced multi-scale recurrent attention network for saliency detection. It achieves dramatic performance especially in complex scenarios. There are three main contributions of our network that are experimentally demonstrated to have significant practical merits. First, we design an effective depth refinement block using residual connections to fully extract and fuse multi-level paired complementary cues from RGB and depth streams. Second, depth cues with abundant spatial information are innovatively combined with multi-scale context features for accurately locating salient objects. Third, we boost our model's performance by a novel recurrent attention module inspired by Internal Generative Mechanism of human brain. This module can generate more accurate saliency results via comprehensively learning the internal semantic relation of the fused feature and progressively optimizing local details with memory-oriented scene understanding. In addition, we create a large scale RGB-D dataset containing more complex scenarios, which can contribute to comprehensively evaluating saliency models. Extensive experiments on six public datasets and ours demonstrate that our method can accurately identify salient objects and achieve consistently superior performance over 16 state-of-the-art RGB and RGB-D approaches.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/bdb6e818edae527c2f81186758b9ef96c97c5aba

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Rethinking RGB-D Salient Object Detection: Models, Data Sets, and Large-Scale Benchmarks (2019, methodology): “After this submission, there are many interesting models, such as UCNet [116], DMRA [117], Molf [118], BiANet [119], etc, have been released.”
RGB-D salient object detection: A survey (2020, methodology): “SE [44]) and eight state-of-the-art CNN-based models (DMRA [101], D3Net [34], SSF [167], A2dele [106], S2MA [82], ICNet [71], JL-DCF [42], and UC-Net [160]).”
Visual Saliency Transformer (2021, methodology): “Some works introduced the attention mechanism to learn more discriminative features, including spatial and channel attention [52, 80, 16, 7] or pixel-wise contextual attention [37].”
Suppress and Balance: A Simple Gated Network for Salient Object Detection (2020, methodology): “etwork pipeline. A.2 RGB-D Salient object detection Dataset. There are ve main RGB-D SOD datasets which are NJUD [23], RGBD135 [9] NLPR [34], SSD [74] and SIP [16]. We adopt the same splitting way as [3,5,18,71,36] to guarantee a fair comparison. We split 1,485 samples from NJUD and 700 samples from”
JL-DCF: Joint Learning and Densely-Cooperative Fusion Framework for RGB-D Salient Object Detection (2020, methodology): “Notable performance gains of JL-DCF over existing and recently proposed techniques, like CPFP [77], D3Net [18] and DMRA [46], can be seen in all four metrics.”
SwinNet: Swin Transformer Drives Edge-Aware RGB-D and RGB-T Salient Object Detection (2022, methodology): “As can be clearly observed from figure that our curves are significant better than the others on NLPR, NJU2K, STERE, SIP and DUT datasets, slightly on DES dataset.”
UC-Net: Uncertainty Inspired RGB-D Saliency Detection via Conditional Variational Autoencoders (2020, methodology): “8 show that our method can not only produce high accuracy predictions (compared with DMRA [61]), but also diverse predictions (compared with M-head based and MC-dropout based models) for images with complex background (image in the first and last rows).”
BBS-Net: RGB-D Salient Object Detection with a Bifurcated Backbone Strategy Network (2020, methodology): “To the best of our knowledge, BBS-Net is the ﬁrst work to explore the cascaded reﬁnement mechanism for the RGB-D SOD task.”
Generating Diverse and Natural 3D Human Motions from Text (2022, background): “…has been fundamentally changed in the deep learning era, where we witness significant performance boosts by adopting a variety of powerful techniques including RNNs [30,40], transformer [18], attentive context modeling [41], memory network [28, 46], GANs [26], and reinforcement learning [19,27].”