[1]李沼洁,郭家毅,杨英东,等.融合Transformer和CNN的伪装目标检测[J].福建理工大学学报,2025,23(06):557-563.[doi:10.3969/j.issn.2097-3853.2025.06.007]
　LI Zhaojie,GUO Jiayi,YANG Yingdong,et al.Camouflaged object detection by fusing Transformer and CNN[J].Journal of Fujian University of Technology;,2025,23(06):557-563.[doi:10.3969/j.issn.2097-3853.2025.06.007]
点击复制

融合Transformer和CNN的伪装目标检测()
分享到：

《福建理工大学学报》[ISSN:2097-3853/CN:35-1351/Z]

卷:
第23卷
期数:
2025年06期
页码:
557-563
栏目:
出版日期:
2025-12-25

文章信息/Info

Title:
Camouflaged object detection by fusing Transformer and CNN
作者:
李沼洁; 郭家毅; 杨英东; 毛国君
厦门市公安局
Author(s):
LI Zhaojie;  GUO Jiayi;  YANG Yingdong;  MAO Guojun
Xiamen Public Security Bureau
关键词:
伪装目标检测; 双分支; 边界感知; 融合细化
Keywords:
camouflaged object detection; dual-branch; boundary perception; fusion refinement
分类号:
TP391
DOI:
10.3969/j.issn.2097-3853.2025.06.007
文献标志码:
A
摘要:
针对复杂伪装场景下大多数方法难以有效定位、分割伪装目标等问题，提出融合Transformer和卷积神经网络的伪装目标检测方法。首先，为增强伪装目标特征表达能力，提出了双分支编码-解码结构；其次，针对目标定位易出现偏差的问题，设计了双分支边界感知模块；为缓解伪装目标检测过程中出现误检和漏检等问题，构建了一个交互式融合细化机制，利用特征分组技术细化伪装目标的纹理，提升复杂场景下模型检测精度。在3个公开数据集上进行了测试，结果表明所提方法表现优异；特别是，相比于经典的ZoomNet算法，所提方法在CAMO数据集上平均绝对误差降低了4.76%，结构相似性度量S-measure提高1.59%。
Abstract:
Addressing the challenges of effectively locating and segmenting camouflaged objects in complex camouflage scenarios, where most methods falter, a Transformer and CNN fusion for camouflaged object detection is proposed. Firstly, to enhance the feature representation capability of camouflaged objects, a dual-branch encoder-decoder structure is introduced. Secondly, to address the problem of easy deviation in object localization, a dual-branch boundary perception module is designed；to alleviate the problems of false detection and missed detection in detecting camouflaged objects, an interactive fusion refinement mechanism is constructed. By using feature grouping techniques to refine the texture of camouflaged objects, the detection accuracy of models in complex scenes is improved. The model was tested on three challenging datasets, and the results show that the proposed method performs well. In particular, compared with the classic ZoomNet algorithm, the proposed method has reduced the MAE by 4.76% and the S-measure is increased by 1.59% on the CAMO dataset.

参考文献/References:

［1］ HE J Y，WU X，JIANG Y G，et al. Hookworm detection in wireless capsule endoscopy images with deep learning［J］. IEEE Transactions on Image Processing，2018，27（5）：2379-2392.［2］ FAN D P，ZHOU T，JI G P，et al. Infnet：automatic COVID-19 lung infection segmentation from CT images［J］. IEEE Transactions on Medical Imaging，2020，39（8）：2626-2637.［3］ MISHRA B，GARG D，NARANG P，et al. Dronesurveillance for search and rescue in natural disaster［J］. Computer Communications，2020，156：1-10.［4］ CHU H K，HSU W H，MITRA N J，et al. Camouflage images［J］. ACM Transactions on Graphics，2010，29（4）：1-8.［5］ WANG K, Bi H, 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, 2021, 69(5):5364-5374.［6］ REN J, HU X, ZHU L, et al. Deep textureaware features for camouflaged object detection［J］. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 33(3): 1157-1167.［7］ 李建东,王岩,曲海成.基于Transformer的跨尺度交互学习伪装目标检测［J］.计算机系统应用,2024,33(2):115-124.［8］ YANG F, ZHAI Q, LI X, et al. Uncertaintyguided transformer reasoning for camouflaged object detection［C］∥Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 4146-4155.［9］ BARLOW H B，BLAKEMORE C，PETTIGREW J D. The neural mechanism of binocular depth discrimination［J］. The Journal of Physiology，1967，193（2）：327-342.［10］ SUN Y J，WANG S，CHEN C，et al. Boundaryguided camouflaged object detection［EB/OL］(2022-07-02)［2024-10-07］. https：∥arxiv.org/abs/2207.00794v1.［11］ 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.［12］ FAN D P, JI G P, SUN G, et al. Camouflaged object detection［C］∥Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 2777-2787.［13］ LYU Y Q，ZHANG J，DAI Y C，et al. Simultaneously localize，segment and rank the camouflaged objects［C］∥2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Nashville，TN，USA: IEEE，2021：11586-11596.［14］ SUN Y J，CHEN G，ZHOU T，et al. Contextaware crosslevel fusion network for camouflaged object detection［EB/OL］(2021-05-26)［2024-10-07］. https：∥arxiv.org/abs/2105.12555v1.［15］ ZHOU T，ZHOU Y，GONG C，et al. Feature aggregation and propagation network for camouflaged object detection［J］. IEEE Trans Image Process，2022，31：7036-7047.［16］ LIU J W，ZHANG J，BARNES N. Modeling aleatoric uncertainty for camouflaged object detection［C］∥2022 IEEE/CVF Winter Conference on Applications of Computer Vision （WACV）. Waikoloa，HI，USA: IEEE，2022：2613-2622.［17］ ZHANG M，XU S，PIAO Y R，et al. PreyNet：preying on camouflaged objects［C］∥Proceedings of the 30th ACM International Conference on Multimedia. Lisboa, Portugal: ACM，2022：5323-5332.［18］ KHAN A，KHAN M，GUEAIEB W，et al. CamoFocus：enhancing camouflage object detection with splitfeature focal modulation and context refinement［C］∥2024 IEEE/CVF Winter Conference on Applications of Computer Vision （WACV）. Waikoloa，HI，USA: IEEE，2024：1423-1432.［19］ 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］∥2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. New Orleans，LA，USA: IEEE，2022：2150-2160.

更新日期/Last Update: 2025-12-25