Effective full-scale detection for salient object based on condensing-and-filtering network
Effective full-scale detection for salient object based on condensing-and-filtering network
复制标题
基于压缩滤波网络的显着目标有效全尺度检测
DOI：
10.1016/j.patcog.2022.108904
复制
发表时间：
2022-07
影响因子：
8
通讯作者：
Qi Tian
中科院分区：
文献类型：
--
作者：
Xinyu Yan;Meijun Sun;Yahong Han;Zheng Wang;Qi Tian
·在显著对象检测领域中仍然存在两个挑战：1）缺乏从不同编码器级别的多个视角提取的丰富特征导致遗漏具有不同尺度的显著对象。2)在解码过程中，多层次特征的无效融合稀释了显著性特征，这破坏了预测图的纯度。·针对上述两个问题，本文提出了一种压缩过滤网络（CFNet），其中提出了显著性金字塔压缩模块（SPCM）和显著性过滤模块（SFM），以实现对显著对象的有效全尺度检测。·实验结果表明，该方法优于23个国家的最先进的方法，具有实时速度和相当大的计算在5个基准数据集。随着深度学习的发展，显著目标检测方法取得了很大的进步。然而，仍然存在两个挑战：1）缺乏从不同编码器级别的多个视角提取的丰富特征，导致省略具有不同尺度的显著对象。2)在解码过程中，多层次特征的无效融合稀释了显著性特征，这破坏了预测图的纯度。本文设计了一个压缩过滤网络（CFNet），提出了显著性金字塔压缩模块（SPCM）和显著性过滤模块（SFM）来分别解决上述两个问题。具体而言，SPCM引入金字塔卷积作为基本单元，以在编码器的每个级别从全局和局部角度浓缩全尺度特征。SFM配备了巧妙的“漏斗”结构，有效过滤多层次特征并补充细节，使特征的融合更加稳健。这两个模块相互补充，使得全尺度特征可以有效地用于预测显著对象。在五个基准数据集上的大量实验结果表明，我们的方法与最先进的方法相比表现良好，并且在速度（16.18ms）和FLOP（21.19G）方面也表现出优越性。同时，我们将CFNet扩展到RGB-D显著对象检测任务中，并取得了更好的结果，进一步证明了其有效性。代码将被提供。
• There are still two challenges in the field of salient object detection: 1) The lack of rich features extracted from multiple perspectives at different encoder levels results in the omission of salient objects with varying scales. 2) The ineffective fusion of multi-level features during decoding dilutes the saliency features, which destroys the purity of the predicted maps. • To solve the above two problems, we propose a Condensing-and-Filtering Network (CFNet), in which a saliency pyramid condensing module (SPCM) and a saliency filtering module (SFM) are proposed to achieve an effective full-scale detection for salient objects. • Experimental results demonstrate that the proposed method outperforms 23 state-of-the-art methods with a real-time speed and considerable computation on five benchmark datasets. With the development of deep learning, salient object detection methods have made great progress. However, there are still two challenges: 1) The lack of rich features extracted from multiple perspectives at different encoder levels results in the omission of salient objects with varying scales. 2) The ineffective fusion of multi-level features during decoding dilutes the saliency features, which destroys the purity of the predicted maps. In this paper, we design a Condensing-and-Filtering Network (CFNet), in which a saliency pyramid condensing module (SPCM) and a saliency filtering module (SFM) are proposed to solve the above two problems respectively. Specifically, SPCM introduces pyramid convolution as the basic unit to condense full-scale features from global and local perspectives at each level of the encoder. SFM is equipped with an ingenious ‘funnel structure to effectively filter multi-level features and supplement details, which makes the fusion of features more robust. The two modules complement each other, so that the full-scale features can be used effectively to predict salient objects. Extensive experimental results on five benchmark datasets demonstrate that our method performs favourably against the state-of-the-art approaches, and also shows superiority in terms of speed (16.18ms) and FLOPs (21.19G). Meanwhile, we extend our CFNet to the task of RGB-D salient object detection and achieve better results, which further demonstrate its effectiveness. The code will be made available.
登录
查看更多内容
影响因子：
8
作者：
Yang Liu;Xinbo Gao;Quanxue Gao;Jungong Han;Ling Shao
通讯作者：
Ling Shao
DOI：
10.1109/tcsvt.2021.3069848
发表时间：
2021-03
影响因子：
8.4
作者：
Haiyang Mei;Liu Yuanyuan;Ziqi Wei;Li Zhu;Yuxin Wang;D. Zhou;Qiang Zhang;Xin Yang
通讯作者：
Haiyang Mei;Liu Yuanyuan;Ziqi Wei;Li Zhu;Yuxin Wang;D. Zhou;Qiang Zhang;Xin Yang
DOI：
10.1109/cvpr.2009.5206596
发表时间：
2009-06
期刊：
2009 IEEE Conference on Computer Vision and Pattern Recognition
影响因子：
--
作者：
R. Achanta;S. Hemami;F. Estrada;S. Süsstrunk
通讯作者：
R. Achanta;S. Hemami;F. Estrada;S. Süsstrunk
DOI：
10.1109/wacv.2018.00171
发表时间：
2017-11
期刊：
2018 IEEE Winter Conference on Applications of Computer Vision (WACV)
影响因子：
--
作者：
Prakhar Gupta;Shubh Gupta;Ajaykrishnan Jayagopal;Sourav Pal;R. Sinha
通讯作者：
Prakhar Gupta;Shubh Gupta;Ajaykrishnan Jayagopal;Sourav Pal;R. Sinha
DOI：
10.1145/3394171.3413969
发表时间：
2020-10
期刊：
Proceedings of the 28th ACM International Conference on Multimedia
影响因子：
--
作者：
Miao Zhang;Yu Zhang;Yongri Piao;Beiqi Hu;Huchuan Lu
通讯作者：
Miao Zhang;Yu Zhang;Yongri Piao;Beiqi Hu;Huchuan Lu