《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (6): 1910-1918.DOI: 10.11772/j.issn.1001-9081.2022050706
所属专题: 多媒体计算与计算机仿真
王强1,2, 黄小明1,2, 佟强1,2, 刘秀磊1,2()
收稿日期:
2022-05-18
修回日期:
2023-01-04
接受日期:
2023-01-10
发布日期:
2023-06-08
出版日期:
2023-06-10
通讯作者:
刘秀磊
作者简介:
王强(1996—),男,安徽潜山人,硕士研究生,主要研究方向:机器学习、图像识别基金资助:
Qiang WANG1,2, Xiaoming HUANG1,2, Qiang TONG1,2, Xiulei LIU1,2()
Received:
2022-05-18
Revised:
2023-01-04
Accepted:
2023-01-10
Online:
2023-06-08
Published:
2023-06-10
Contact:
Xiulei LIU
About author:
WANG Qiang, born in 1996, M. S. candidate. His research interests include machine learning, image recognition.Supported by:
摘要:
针对以往的弱监督显著性目标检测算法存在的显著目标定位不准确问题,提出一种基于边界框标注的弱监督显著目标检测算法。所提算法利用图像中所有目标的最小外接矩形框,即边界框,作为监督信息。首先基于边界框标注和GrabCut算法生成初始显著图;然后在此基础上设计了一个缺失修正模块,以得到优化后的显著图;最后结合传统方法和深度学习方法各自的优势,将优化后的显著图作为伪真值,通过神经网络学习一个显著性目标检测模型。在4个公开数据集上与6种无监督、4种弱监督的显著性检测算法进行比较的实验结果显示,所提算法在所有数据集上的最大F度量值(Max-F)和平均绝对误差(MAE)均明显优于对比算法:与同样基于边界框标注的弱监督方法SBB(Saliency Bounding Boxes)相比,所提算法的标注方法更简单,在ECSSD、DUTS-TE、HKU-IS、DUT-OMRON等4个数据集上进行实验,Max-F分别提高了1.82%、4.00%、1.27%和5.33%,MAE分别降低了13.89%、15.07%、8.77%和13.33%。可见,所提算法是一种具有良好检测性能的弱监督显著目标检测算法。
中图分类号:
王强, 黄小明, 佟强, 刘秀磊. 基于边界框标注的弱监督显著性目标检测算法[J]. 计算机应用, 2023, 43(6): 1910-1918.
Qiang WANG, Xiaoming HUANG, Qiang TONG, Xiulei LIU. Weakly supervised salient object detection algorithm based on bounding box annotation[J]. Journal of Computer Applications, 2023, 43(6): 1910-1918.
算法 | 监督方式 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | ||
MR[ | 无监督 | 0.690 | 0.186 | 0.510 | 0.189 | 0.655 | 0.174 | 0.577 | 0.194 |
DSR[ | 无监督 | 0.676 | 0.179 | 0.506 | 0.163 | 0.677 | 0.149 | 0.536 | 0.145 |
HS[ | 无监督 | 0.627 | 0.229 | 0.460 | 0.258 | 0.623 | 0.223 | 0.507 | 0.237 |
BSCA[ | 无监督 | 0.707 | 0.185 | 0.500 | 0.197 | 0.654 | 0.175 | 0.509 | 0.190 |
MB+[ | 无监督 | 0.697 | 0.174 | 0.528 | 0.179 | 0.678 | 0.151 | 0.531 | 0.167 |
MST[ | 无监督 | 0.693 | 0.151 | 0.540 | 0.156 | 0.680 | 0.131 | 0.542 | 0.149 |
ASMO[ | 弱监督(类别标注) | 0.810 | 0.114 | 0.625 | 0.123 | 0.821 | 0.091 | 0.633 | 0.100 |
WSS[ | 弱监督(类别标注) | 0.828 | 0.105 | 0.657 | 0.106 | 0.821 | 0.081 | 0.611 | 0.111 |
MSW[ | 弱监督(类别+标题标注) | 0.846 | 0.096 | 0.704 | 0.097 | 0.823 | 0.086 | 0.619 | 0.109 |
SBB[ | 弱监督(边界框标注) | 0.878 | 0.072 | 0.775 | 0.073 | 0.869 | 0.057 | 0.751 | 0.075 |
本文算法 | 弱监督(边界框标注) | 0.894 | 0.062 | 0.806 | 0.062 | 0.880 | 0.052 | 0.791 | 0.065 |
表1 本文算法与其他算法的性能定量比较
Tab. 1 Quantitative comparison of the proposed algorithm and other algorithms
算法 | 监督方式 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | ||
MR[ | 无监督 | 0.690 | 0.186 | 0.510 | 0.189 | 0.655 | 0.174 | 0.577 | 0.194 |
DSR[ | 无监督 | 0.676 | 0.179 | 0.506 | 0.163 | 0.677 | 0.149 | 0.536 | 0.145 |
HS[ | 无监督 | 0.627 | 0.229 | 0.460 | 0.258 | 0.623 | 0.223 | 0.507 | 0.237 |
BSCA[ | 无监督 | 0.707 | 0.185 | 0.500 | 0.197 | 0.654 | 0.175 | 0.509 | 0.190 |
MB+[ | 无监督 | 0.697 | 0.174 | 0.528 | 0.179 | 0.678 | 0.151 | 0.531 | 0.167 |
MST[ | 无监督 | 0.693 | 0.151 | 0.540 | 0.156 | 0.680 | 0.131 | 0.542 | 0.149 |
ASMO[ | 弱监督(类别标注) | 0.810 | 0.114 | 0.625 | 0.123 | 0.821 | 0.091 | 0.633 | 0.100 |
WSS[ | 弱监督(类别标注) | 0.828 | 0.105 | 0.657 | 0.106 | 0.821 | 0.081 | 0.611 | 0.111 |
MSW[ | 弱监督(类别+标题标注) | 0.846 | 0.096 | 0.704 | 0.097 | 0.823 | 0.086 | 0.619 | 0.109 |
SBB[ | 弱监督(边界框标注) | 0.878 | 0.072 | 0.775 | 0.073 | 0.869 | 0.057 | 0.751 | 0.075 |
本文算法 | 弱监督(边界框标注) | 0.894 | 0.062 | 0.806 | 0.062 | 0.880 | 0.052 | 0.791 | 0.065 |
方法 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | |
GrabCut生成的初始显著图 | 0.892 7 | 0.056 4 | 0.814 0 | 0.058 7 | 0.855 7 | 0.061 4 | 0.870 5 | 0.040 5 |
GrabCut +四周调整的显著图 | 0.893 5 | 0.054 6 | 0.817 0 | 0.057 4 | 0.857 0 | 0.059 9 | 0.871 7 | 0.039 7 |
GrabCut +四周中部调整的显著图 | 0.894 1 | 0.054 4 | 0.817 2 | 0.057 4 | 0.856 9 | 0.060 0 | 0.872 0 | 0.039 6 |
GrabCut +全部后处理的显著图 | 0.894 3 | 0.054 2 | 0.817 5 | 0.057 3 | 0.857 1 | 0.059 9 | 0.872 2 | 0.039 5 |
表2 本文算法显著图生成过程中的结果的定量比较
Tab. 2 Quantitative comparison of results in saliency map generation process of the proposed algorithm
方法 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | |
GrabCut生成的初始显著图 | 0.892 7 | 0.056 4 | 0.814 0 | 0.058 7 | 0.855 7 | 0.061 4 | 0.870 5 | 0.040 5 |
GrabCut +四周调整的显著图 | 0.893 5 | 0.054 6 | 0.817 0 | 0.057 4 | 0.857 0 | 0.059 9 | 0.871 7 | 0.039 7 |
GrabCut +四周中部调整的显著图 | 0.894 1 | 0.054 4 | 0.817 2 | 0.057 4 | 0.856 9 | 0.060 0 | 0.872 0 | 0.039 6 |
GrabCut +全部后处理的显著图 | 0.894 3 | 0.054 2 | 0.817 5 | 0.057 3 | 0.857 1 | 0.059 9 | 0.872 2 | 0.039 5 |
方法 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | |
GrabCut生成的初始显著图 | 0.886 5 | 0.068 9 | 0.803 9 | 0.064 3 | 0.874 9 | 0.054 7 | 0.781 7 | 0.067 2 |
GrabCut +四周调整的显著图 | 0.890 6 | 0.064 5 | 0.804 0 | 0.063 2 | 0.876 0 | 0.053 7 | 0.793 2 | 0.064 3 |
GrabCut +四周、中部调整的显著图 | 0.893 8 | 0.063 2 | 0.807 4 | 0.062 9 | 0.875 4 | 0.053 6 | 0.794 2 | 0.064 6 |
GrabCut +全部后处理的显著图 | 0.893 6 | 0.062 1 | 0.805 7 | 0.061 9 | 0.879 6 | 0.052 0 | 0.791 2 | 0.065 1 |
表3 不同显著图后处理结果作为伪真值输入给PoolNet训练模型的性能定量比较
Tab. 3 Quantitative comparison of performance of PoolNet trained models with input of different saliency map post-processing results as pseudo ground-truth
方法 | ECSSD | DUTS-TE | HKU-IS | DUT-OMRON | ||||
---|---|---|---|---|---|---|---|---|
Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | Max-F↑ | MAE↓ | |
GrabCut生成的初始显著图 | 0.886 5 | 0.068 9 | 0.803 9 | 0.064 3 | 0.874 9 | 0.054 7 | 0.781 7 | 0.067 2 |
GrabCut +四周调整的显著图 | 0.890 6 | 0.064 5 | 0.804 0 | 0.063 2 | 0.876 0 | 0.053 7 | 0.793 2 | 0.064 3 |
GrabCut +四周、中部调整的显著图 | 0.893 8 | 0.063 2 | 0.807 4 | 0.062 9 | 0.875 4 | 0.053 6 | 0.794 2 | 0.064 6 |
GrabCut +全部后处理的显著图 | 0.893 6 | 0.062 1 | 0.805 7 | 0.061 9 | 0.879 6 | 0.052 0 | 0.791 2 | 0.065 1 |
1 | 张文达,许悦雷,倪嘉成,等. 基于多尺度分块卷积神经网络的图像目标识别算法[J]. 计算机应用, 2016, 36(4): 1033-1038. 10.11772/j.issn.1001-9081.2016.04.1033 |
ZHANG W D, XU Y L, NI J C, et al. Image target recognition method based on multi-scale block convolutional neural network[J]. Journal of Computer Applications, 2016, 36(4): 1033-1038. 10.11772/j.issn.1001-9081.2016.04.1033 | |
2 | YANG X Y, QIAN X M, XUE Y. Scalable mobile image retrieval by exploring contextual saliency [J]. IEEE Transactions on Image Processing, 2015, 24(6): 1709-1721. 10.1109/tip.2015.2411433 |
3 | SU Y Y, ZHAO Q J, ZHAO L J, et al. Abrupt motion tracking using a visual saliency embedded particle filter[J]. Pattern Recognition, 2014, 47(5): 1826-1834. 10.1016/j.patcog.2013.11.028 |
4 | ACHANTA R, HEMAMI S, ESTRADA F, et al. Frequency-tuned salient region detection[C]// Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2009: 1597-1604. 10.1109/cvpr.2009.5206596 |
5 | GOFERMAN S, ZELNIK-MANOR L, TAL A. Context-aware saliency detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(10): 1915-1926. 10.1109/tpami.2011.272 |
6 | ZHANG J M, SCLAROFF S, LIN Z, et al. Minimum barrier salient object detection at 80 FPS[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 1404-1412. 10.1109/iccv.2015.165 |
7 | ACHANTA R, ESTRADA F, WILS P, et al. Salient region detection and segmentation[C]// Proceedings of the 2008 International Conference on Computer Vision Systems, LNCS 5008. Berlin: Springer, 2008: 66-75. |
8 | VALENTI R, SEBE N, GEVERS T. Image saliency by isocentric curvedness and color[C]// Proceedings of the IEEE 12th International Conference on Computer Vision. Piscataway: IEEE, 2009: 2185-2192. 10.1109/iccv.2009.5459240 |
9 | ZHANG J, ZHANG T, DAI Y C, et al. Deep unsupervised saliency detection: a multiple noisy labeling perspective[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 9029-9038. 10.1109/cvpr.2018.00941 |
10 | CHENG M M, MITRA N J, HUANG X L, et al. Global contrast based salient region detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(3): 569-582. 10.1109/tpami.2014.2345401 |
11 | 李小雨,房体育,夏英杰,等. 基于图割精细化和可微分聚类的无监督显著性目标检测[J]. 计算机应用, 2021, 41(12): 3571-3577. 10.11772/j.issn.1001-9081.2021061054 |
LI X Y, FANG T Y, XIA Y J, et al. Unsupervised salient object detection based on graph cut refinement and differentiable clustering [J]. Journal of Computer Applications, 2021, 41(12): 3571-3577. 10.11772/j.issn.1001-9081.2021061054 | |
12 | YANG C, ZHANG L H, LU H C, et al. Saliency detection via graph-based manifold ranking [C]// Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2013: 3166-3173. 10.1109/cvpr.2013.407 |
13 | LI X H, LU H C, ZHANG L H, et al. Saliency detection via dense and sparse reconstruction[C]// Proceedings of the 2013 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2013: 2976-2983. 10.1109/iccv.2013.370 |
14 | YAN Q, XU L, SHI J P, et al. Hierarchical saliency detection [C]// Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2013: 1155-1162. 10.1109/cvpr.2013.153 |
15 | QIN Y, LU H C, XU Y Q, et al. Saliency detection via cellular automata[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 110-119. 10.1109/cvpr.2015.7298606 |
16 | TU W C, HE S F, YANG Q X, et al. Real-time salient object detection with a minimum spanning tree [C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 2334-2342. 10.1109/cvpr.2016.256 |
17 | 赵恒,安维胜,付为刚. 深度导向显著性检测算法[J]. 计算机应用, 2019, 39(1): 143-147. 10.11772/j.issn.1001-9081.2018061194 |
ZHAO H, AN W S, FU W G. Saliency detection algorithm of deep guidance[J]. Journal of Computer Applications, 2019, 39(1): 143-147. 10.11772/j.issn.1001-9081.2018061194 | |
18 | 后云龙,朱磊,陈琴,等. 基于高斯差分特征网络的显著目标检测[J]. 计算机应用, 2021, 41(3): 706-713. 10.11772/j.issn.1001-9081.2020060957 |
HOU Y L, ZHU L, CHEN Q, et al. Salient object detection based on difference of Gaussian feature network [J]. Journal of Computer Applications, 2021, 41(3): 706-713. 10.11772/j.issn.1001-9081.2020060957 | |
19 | 温静,宋建伟. 基于多级全局信息传递模型的视觉显著性检测[J]. 计算机应用, 2021, 41(1): 208-214. 10.11772/j.issn.1001-9081.2020060968 |
WEN J, SONG J W. Visual saliency detection based on multi-level global information propagation model [J]. Journal of Computer Applications, 2021, 41(1): 208-214. 10.11772/j.issn.1001-9081.2020060968 | |
20 | LI G B, XIE Y, LIN L. Weakly supervised salient object detection using image labels [C]// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2018: 7024-7031. 10.1609/aaai.v32i1.12308 |
21 | 谭台哲,轩康西,曾群生. 基于图像级标签及超像素块的弱监督显著性检测[J]. 计算机应用研究, 2020, 37(2): 601-605. 10.19734/j.issn.1001-3695.2018.06.0576 |
TAN T Z, XUAN K X, ZENG Q S. Supervised significant detection based on image level labels and superpixel blocks [J]. Application Research of Computers, 2020, 37(2): 601-605. 10.19734/j.issn.1001-3695.2018.06.0576 | |
22 | WANG L J, LU H C, WANG Y F, et al. Learning to detect salient objects with image-level supervision [C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 3796-3805. 10.1109/cvpr.2017.404 |
23 | ZENG Y, ZHUGE Y Z, LU H C, et al. Multi-source weak supervision for saliency detection[C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 6067-6076. 10.1109/cvpr.2019.00623 |
24 | LIU Y X, WANG P J, CAO Y, et al. Weakly-supervised salient object detection with saliency bounding boxes[J]. IEEE Transactions on Image Processing, 2021, 30: 4423-4435. 10.1109/tip.2021.3071691 |
25 | DAI J F, HE K M, SUN J. BoxSup: exploiting bounding boxes to supervise convolutional networks for semantic segmentation [C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 1635-1643. 10.1109/iccv.2015.191 |
26 | KHOREVA A, BENENSON R, HOSANG J, et al. Simple does it: weakly supervised instance and semantic segmentation[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 876-885. 10.1109/cvpr.2017.181 |
27 | ROTHER C, KOLMOGOROV V, BLAKE A. “GrabCut”: interactive foreground extraction using iterated graph cuts[J]. ACM Transactions on Graphics, 2004, 23(3): 309-314. 10.1145/1015706.1015720 |
28 | LIU J J, HOU Q B, CHENG M M, et al. A simple pooling-based design for real-time salient object detection [C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 3912-3921. 10.1109/cvpr.2019.00404 |
29 | BOYKOV Y, JOLLY M P. Interactive graph cuts for optimal boundary & region segmentation of objects in N-D images [C]// Proceedings 8th IEEE International Conference on Computer Vision. Piscataway: IEEE, 2001,1: 105-112. |
30 | LI G B, YU Y Z. Visual saliency based on multiscale deep features[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 5455-5463. 10.1109/cvpr.2015.7299184 |
[1] | 徐松, 张文博, 王一帆. 基于时空信息的轻量视频显著性目标检测网络[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2192-2199. |
[2] | 朱子蒙, 李志新, 郇战, 陈瑛, 梁久祯. 基于三元中心引导的弱监督视频异常检测[J]. 《计算机应用》唯一官方网站, 2024, 44(5): 1452-1457. |
[3] | 党伟超, 张磊, 高改梅, 刘春霞. 融合片段对比学习的弱监督动作定位方法[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 548-555. |
[4] | 林呈宇, 王雷, 薛聪. 标签语义增强的弱监督文本分类模型[J]. 《计算机应用》唯一官方网站, 2023, 43(2): 335-342. |
[5] | 胡聪, 华钢. 基于注意力机制的弱监督动作定位方法[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 960-967. |
[6] | 陈权, 李莉, 陈永乐, 段跃兴. 面向深度学习可解释性的对抗攻击算法[J]. 《计算机应用》唯一官方网站, 2022, 42(2): 510-518. |
[7] | 罗萍, 丁玲, 杨雪, 向阳. 基于数据增强和弱监督对抗训练的中文事件检测[J]. 《计算机应用》唯一官方网站, 2022, 42(10): 2990-2995. |
[8] | 邓爽, 何小海, 卿粼波, 陈洪刚, 滕奇志. 基于改进VGG网络的弱监督细粒度阿尔兹海默症分类方法[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 302-309. |
[9] | 陆鑫伟, 余鹏飞, 李海燕, 李红松, 丁文谦. 基于注意力自身线性融合的弱监督细粒度图像分类算法[J]. 计算机应用, 2021, 41(5): 1319-1325. |
[10] | 李小雨, 房体育, 夏英杰, 李金屏. 基于图割精细化和可微分聚类的无监督显著性目标检测[J]. 《计算机应用》唯一官方网站, 2021, 41(12): 3571-3577. |
[11] | 汪虹余, 张彧, 杨恒, 穆楠. 基于蚁群优化算法的弱光图像显著性目标检测[J]. 计算机应用, 2021, 41(10): 2970-2978. |
[12] | 边小勇, 江沛龄, 赵敏, 丁胜, 张晓龙. 基于多分支神经网络模型的弱监督细粒度图像分类方法[J]. 计算机应用, 2020, 40(5): 1295-1300. |
[13] | 周健, 黄章进. 基于改进三维形变模型的三维人脸重建和密集人脸对齐方法[J]. 计算机应用, 2020, 40(11): 3306-3313. |
[14] | 严经纬, 李强, 王春茂, 谢迪, 王保青, 戴骏. 面部运动单元检测研究综述[J]. 计算机应用, 2020, 40(1): 8-15. |
[15] | 丁英姿, 丁香乾, 郭保琪. 基于弱监督的改进型GoogLeNet在DR检测中的应用[J]. 计算机应用, 2019, 39(8): 2484-2488. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||