融合多尺度交叉注意力和边缘感知的伪装目标检测

doi:10.3778/j.issn.1002-8331.2401-0200

摘要/Abstract

摘要： 针对当前伪装目标检测算法无法准确、完整地检测出目标对象和其边缘的问题，提出了一种融合多尺度交叉注意力和边缘感知的伪装目标检测网络（multi-scale cross attention and edge perception network，MAEP-Net）。利用Res2Net-50提取图像的原始特征，并采用融合了多尺度交叉注意力的特征金字塔结构从通道、空间两个维度挖掘目标位置信息和凸显伪装目标区域特征；使用定位模块对目标的大致位置进行准确定位；边缘感知模块抑制低级特征中背景的噪声，融合边缘特征以获取更多的边缘细节信息；细化模块通过注意力机制分别从前景和背景两个方向关注目标线索，利用边缘先验、语义先验、领域先验、区域先验知识进一步细化目标结构和边缘轮廓。在3个公开数据集上的实验表明，所提算法相较于12种主流算法在4个客观评价指标上均取得了最优表现，尤其是在COD10K数据集上所提算法的加权平均值F-measure和平均绝对误差（mean absolute error，MAE）分别达到0.797和0.031。由此可见，所提算法在COD任务上具有较好的检测效果。

关键词: 多尺度交叉注意力, 边缘感知, 伪装目标检测, 特征金字塔结构

Abstract: Aiming at the problem that current camouflaged object detection algorithms being unable accurately and completely detect target objects and their edges, a camouflaged object detection network integrating multi-scale cross attention and edge perception (MAEP-Net) is proposed. Use Res2Net-50 to extract the original features of the image, use a feature pyramid structure that integrates multi-scale cross attention to mine target position information and highlight disguised target area features from both channel and spatial dimensions. Use a positioning module to accurately locate the approximate position of the target. The edge perception module suppresses background noise in low-level features and fuses edge features to obtain more edge detail information. The final refining module concentrates on target clues from both foreground and background directions through attention mechanisms, and further refines the target structure and edge contours by using edge prior, semantic prior, domain prior, and region prior knowledge. Experimental results on 3 public datasets have shown that the proposed algorithm outperforms 12 mainstream algorithms in all 4 objective evaluation metrics, especially on the COD10K dataset where the weighted average F-measure and mean absolute error (MAE) of the proposed algorithm reach 0.797 and 0.031 respectively. It follows that the proposed algorithm has good detection performance in COD tasks.

Key words: multi-scale cross attention, edge perception, camouflaged object detection, feature pyramid structure

郝子强, 张庆宝, 赵世豪, 王焯豪, 詹伟达. 融合多尺度交叉注意力和边缘感知的伪装目标检测[J]. 计算机工程与应用, 2025, 61(10): 228-237.

HAO Ziqiang, ZHANG Qingbao, ZHAO Shihao, WANG Zhuohao, ZHAN Weida. Camouflaged Object Detection Based on Multi-Scale Cross Attention and Edge Perception[J]. Computer Engineering and Applications, 2025, 61(10): 228-237.

参考文献

[1] FAN D P, JI G P, SUN G L, et al. Camouflaged object detection[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 2774-2784.
[2] WU Z, SU L, HUANG Q M. Cascaded partial decoder for fast and accurate salient object detection[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 3902-3911.
[3] WANG K, BI H B, ZHANG Y, et al. D2C-net: a dual-branch, dual-guidance and cross-refine network for camouflaged object detection[J]. IEEE Transactions on Industrial Electronics, 2022, 69(5): 5364-5374.
[4] RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]//Proceedings of the 2015 International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241.
[5] SUN Y J, CHEN G, ZHOU T, et al. Context-aware cross-level fusion network for camouflaged object detection[J]. arXiv:2105.12555, 2021.
[6] LIN T Y, DOLLáR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 936-944.
[7] GAO S H, CHENG M M, ZHAO K, et al. Res2Net: a new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(2): 652-662.
[8] 孙备, 党昭洋, 吴鹏, 等. 多尺度互交叉注意力改进的单无人机对地伪装目标检测定位方法[J]. 仪器仪表学报, 2023, 44(6): 54-65.
SUN B, DANG Z Y, WU P, et al. Multi scale cross attention improved method of single unmanned aerial vehicle for ground camouflage target detection and localization[J]. Chinese Journal of Scientific Instrument, 2023, 44(6): 54-65.
[9] SUN Y J, WANG S, CHEN C, et al. Boundary-guided camouflaged object detection[J]. arXiv:2207.00794, 2022.
[10] 詹春兰, 王安志, 王明辉. 基于通道注意力和边缘融合的伪装目标分割方法[J]. 计算机应用, 2023, 43(7): 2166-2172.
ZHAN C L, WANG A Z, WANG M H. Camouflage object segmentation method based on channel attention and edge fusion[J]. Journal of Computer Applications, 2023, 43(7): 2166-2172.
[11] 张冬冬, 王春平, 付强. 基于人眼视觉机制的伪装目标检测网络[J/OL]. 北京航空航天大学学报 [2023-12-28]. https://doi.org/10.13700/j.bh.1001-5965.2023.0511.
ZHANG D D, WANG C P, FU Q. Camouflaged object detection network based on human visual mechanisms[J/OL]. Journal of Beijing University of Aeronautics and Astronautics[2023-12-28]. https://doi.org/10.13700/j.bh.1001-5965.2023.
0511.
[12] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision. Cham: Springer, 2018: 3-19.
[13] WEI J, WANG S H, HUANG Q M. F3Net: fusion, feedback and focus for salient object detection[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7): 12321-12328.
[14] XIE E Z, WANG W J, WANG W H, et al. Segmenting transparent objects in the wild[C]//Proceedings of the 16th European Conference on Computer Vision. Cham: Springer, 2020: 696-711.
[15] LE T N, NGUYEN T V, NIE Z L, et al. Anabranch network for camouflaged object segmentation[J]. Computer Vision and Image Understanding, 2019, 184: 45-56.
[16] LV Y Q, ZHANG J, DAI Y C, et al. Simultaneously localize, segment and rank the camouflaged objects[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 11586-11596.
[17] ZHOU T, ZHOU Y, GONG C, et al. Feature aggregation and propagation network for camouflaged object detection[J]. IEEE Transactions on Image Processing, 2022, 31: 7036-7047.
[18] CHENG M M, FAN D P. Structure-measure: a new way to evaluate foreground maps[J]. International Journal of Computer Vision, 2021, 129(9): 2622-2638.
[19] FAN D P, JI G P, QIN X B, et al. Cognitive vision inspired object segmentation metric and loss function[J]. Scientia Sinica Informationis, 2021, 51(9): 1475.
[20] 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.
[21] PERAZZI F, KR?HENBüHL P, PRITCH Y, et al. Saliency filters: contrast based filtering for salient region detection[C]//Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2012: 733-740.
[22] LI A X, ZHANG J, LV Y Q, et al. Uncertainty-aware joint salient object and camouflaged object detection[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 10066-10076.
[23] ZHAI Q, LI X, YANG F, et al. Mutual graph learning for camouflaged object detection[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 12992-13002.
[24] MEI H Y, JI G P, WEI Z Q, et al. Camouflaged object segmentation with distraction mining[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 8768-8777.
[25] JI G P, ZHU L, ZHUGE M C, et al. Fast camouflaged object detection via edge-based reversible re-calibration network[J]. Pattern Recognition, 2022, 123: 108414.
[26] PANG Y W, ZHAO X Q, XIANG T Z, et al. Zoom in and out: a mixed-scale triplet network for camouflaged object detection[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2022: 2160 -2170.
[27] CHEN T Y, XIAO J, HU X G, et al. Boundary-guided network for camouflaged object detection[J]. Knowledge-Based Systems, 2022, 248: 108901.
[28] 葛斌, 陈宁洁, 夏晨星, 等. 基于全局上下文交互融合的伪装目标检测网络[J]. 控制与决策, 2024, 39(10): 3347-3356.
GE B, CHEN N J, XIA C X, et al. Camouflaged object detection network based on global context interaction fusion[J]. Control and Decision, 2024, 39(10): 3347-3356.
[29] HUANG Z, DAI H, XIANG T Z, et al. Feature shrinkage pyramid for camouflaged object detection with transformers[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023: 5557-5566.