伪装目标检测中的损失函数研究综述

A comprehensive review of loss functions in camouflaged object detection

摘要

基于深度学习的伪装目标检测是计算机视觉领域的重要研究方向,其目标是精准高效地识别与周围环境高度融合的伪装目标。但伪装目标检测面临着前景-背景差异小、边缘模糊、上下文碎片化及预测不确定性等4类核心难题。针对这些问题,研究者们大多从网络架构入手进行改进,而较少从损失函数等非结构化方面进行改进。损失函数是基于深度学习的检测方法的重要组成部分。损失函数的优化设计可以从根本上提升网络的检测效果,并且损失函数与网络结构无关,可以即插即用地应用于各种网络模型和检测任务中。本文从伪装目标检测任务的四大核心挑战出发,系统综述了针对这些挑战的损失函数设计原理、演进及其优缺点,其次在统一的实现与训练协议下,对常用像素级损失进行了跨数据集的定量比较,并评估了若干边界增强组合对结构与边界质量的改进效果。基于理论与实验证据,本文还讨论了损失函数的加权组合策略、工程化参数选择与现实场景下的泛化问题提出了可操作的建议。

Abstract

Camouflaged object detection based on deep learning is an important research direction in the field of computer vision,aiming to accurately and efficiently identify objects that are highly blended with their surrounding environment.However,camouflaged object detection faces four core challenges:minimal foreground/background differences,blurred edges,fragmented context,and prediction uncertainty.To address these issues,most researchers have focused on improving network architectures,while fewer have explored improvements from unstructured aspects such as loss functions.Loss functions are a crucial component of deep learning-based detection methods.The optimized design of loss functions can fundamentally enhance the detection performance of networks.Moreover,loss functions are independent of network structures and can be plug-and-play applied to various network models and detection tasks.Starting from the four core challenges of camouflaged object detection,this paper systematically reviews the design principles,evolution,and advantages and disadvantages of loss functions tailored to these challenges.Furthermore,under a unified implementation and training protocol,it conducts a cross-dataset quantitative comparison of commonly used pixel-level losses and evaluates the improvement effects of several boundary enhancement combinations on structure and boundary quality.Based on theoretical and experimental evidence,the paper also discusses loss function weighting strategies,engineering parameter selection,and generalization issues in real-world scenarios,offering actionable recommendations.

关键词

深度学习 / 伪装目标检测 / 损失函数 / 组合策略

Key words

deep learning / camouflaged object detection / loss function / combination strategy

[1] Fan Dengping,Ji Gepeng,Sun Guolei,et al.Camouflaged object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle,USA:IEEE,2020:2777-2787.
[2] Fan Dengping,Ji Gepeng,Cheng Mingming,et al.Concealed object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2022,44(10):6024-6042.
[3] Huang Kaihong,Li Chunshu,Zhang Jiaqi,et al.Cascade and fusion:a deep learning approach for camouflaged object sensing[J].Sensors,2021,21(16):5455.
[4] 张冬冬,王春平,付强.伪装目标检测研究进展[J].激光杂志,2024,45(3):1-13.
[5] Luo Ziyang,Liu Nian,Zhao Wangbo,et al.Vscode:general visual salient and camouflaged object detection with 2D prompt learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2024:17169-17180.
[6] 赖杰,彭锐晖,孙殿星,等.融合注意力机制与多检测层结构的伪装目标检测[J].中国图象图形学报,2024,29(1):134-146.
[7] 陈乔松,陈鹏昌,李佩,等.基于多尺度特征选择与融合的目标检测方法[J].重庆邮电大学学报(自然科学版),2023,35(2):227-234.
[8] 王毅,瞿治国,孙乐.结合自注意力机制和软阈值降噪的对比联邦学习特征聚合算法[J].重庆邮电大学学报(自然科学版),2024,36(5):966-973.
[9] 吴悦,雒江涛,张攀,等.基于Transformer和位置约束的端到端多目标追踪算法[J].重庆邮电大学学报(自然科学版),2023,35(3):563-570.
[10] Zhong Yijie,Li Bo,Tang Lü,et al.Detecting camouflaged object in frequency domain[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans,USA:IEEE,2022:4504-4513.
[11] Le T N,Cao Yubo,Nguyen T C,et al.Camouflaged instance segmentation in-the-wild:dataset,method,and benchmark suite[J].IEEE Transactions on Image Processing,2021,31:287-300.
[12] Lin Tsung-yi,Goyal P,Girshick R,et al.Focal loss for dense object detection[C]//Proceedings of the IEEE international conference on computer vision.2017:2980-2988.
[13] Wang Anzhi,Ren Chunhong,Zhao Shuang,et al.Attention guided multi-level feature aggregation network for camouflaged object detection[J].Image and Vision Computing,2024,144:104953.
[14] Lü Yunqiu,Zhang Jing,Dai Yuchao,et al.Simultaneously localize,segment and rank the camouflaged objects[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville,USA:IEEE,2021:11591-11601.
[15] Rahman M A,Wang Yang.Optimizing intersection-over-union in deep neural networks for image segmentation[C]//International Symposium on Visual Computing.Cham:Springer International Publishing,2016:234-244.
[16] Zhang Haichao,Qin Can,Yin Yu,et al.Camouflaged image synthesis is all you need to boost camouflaged detection[PP/OL].arXiv (2023-08-13)[2026-06-07].https://arxiv.org/abs/2308.06701.
[17] Rezatofighi H,Tsoi N,Gwak J,et al.Generalized intersection over union:a metric and a loss for bounding box regression[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach,USA:IEEE,2019:658-666.
[18] Zhang Yi,Zhang Jing,Hamidouche W,et al.Predictive uncertainty estimation for camouflaged object detection[J].IEEE Transactions on Image Processing,2023,32:3580-3591.
[19] Milletari F,Navab N,Ahmadi S A.V-Net:fully convolutional neural networks for volumetric medical image segmentation[C]//Proceedings of the 4th International Conference on 3D Vision.Stanford,USA:IEEE,2016:565-571.
[20] Salehi S S M,Erdogmus D,Gholipour A.Tversky loss function for image segmentation using 3D fully convolutional deep networks[C]//International Workshop on Machine Learning in Medical Imaging.Quebec City,Canada:Springer,2017:379-387.
[21] Abraham N,Khan N M.A novel focal tversky loss function with improved attention u-net for lesion segmentation[C]//Proceedings of the 16th International Symposium on Biomedical Imaging. Stanford,USA:IEEE,2019:683-687.
[22] Sudre C H,Li W,Vercauteren T,et al.Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentations[C]//Cardoso M,et al.Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support.Cham:Springer,2017:240-248.
[23] Liu Yu,Li Haihang,Cheng Juan,et al.MSCAF-Net:a general framework for camouflaged object detection via learning multi-scale context-aware features[J].IEEE Transactions on Circuits and Systems for Video Technology,2023,33(9) :4934-4947.
[24] Ren Jingjing,Hu Xiaowei,Zhu Lei,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.
[25] Radford A,Kim J W,Hallacy C,et al.Learning transferable visual models from natural language supervision[C]//Proceedings of the International Conference on Machine Learning,2021:8748-8763.
[26] He Chunming,Li Kai,Zhang Yachao,et al.Camouflaged object detection with feature decomposition and edge reconstruction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Vancouver,Canada:IEEE,2023:22046-22055.
[27] 史彩娟,任弼娟,王子雯,等.基于深度学习的伪装目标检测综述[J].计算机科学与探索,2022,16(12):2734-2751.
[28] Canny J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1986,8(6):679-698.
[29] Ji Gepeng,Fan Dengping,Chou Yucheng,et al.Deep gradient learning for efficient camouflaged object detection[J].Machine Intelligence Research,2023,20(1):92-108.
[30] Ren Chunhong,Wang Anzhi,Yang Chengbang,et al.Frequency domain-based cross-layer feature aggregation network for camouflaged object detection[J].IEEE Signal Processing Letters,2025,32:2005-2009.
[31] Wang Zhou,Bovik A C,Sheikh H R,et al.Image quality assessment:from error visibility to structural similarity[J].IEEE Transactions on Image Processing,2004,13(4):600-612.
[32] Johnson J,Alahi A,Li feifei.Perceptual losses for real-time style transfer and super-resolution[C]//Proceedings of the European Conference on Computer Vision.Amsterdam,Netherlands:Springer,2016:694-711.
[33] Pang Youwei,Zhao Xiaoqi,Xiang Tianzhu,et al.Zoom in and out:a mixed-scale triplet network for camouflaged object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans,USA:IEEE,2022:2160-2170.
[34] Song Ze,Kang Xudong,Wei Xiaohui,et al.FSNet:focus scanning network for camouflaged object detection[J].IEEE Transactions on Image Processing,2023,32(1):2267-2278.
[35] Huang Zhou,Dai Hang,Xiang Tianzhu,et al.Feature shrinkage pyramid for camouflaged object detection with transformers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2023:5557-5566.
[36] Dosovitskiy A,Beyer L,Kolesnikov A,et al.An image is worth 16x16 words:transformers for image recognition at scale[PP/OL].arXiv (2020-10-22)[2026-06-07].https://arxiv.org/abs/2010.11929.
[37] 陈世洁,范李平,余肖生,等.融合上下文感知和背景探索的伪装目标检测方法[J].国外电子测量技术,2024,43(8):17-25.
[38] He Kaiming,Chen Xinlei,Xie Saining,et al.Masked autoencoders are scalable vision learners[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans,USA,2022:16000-16009.
[39] Cong Runmin,Sun Mengyao,Zhang Sanyi,et al.Frequency perception network for camouflaged object detection[C]//Proceedings of the 31st ACM International Conference on Multimedia,2023:1179-1189.
[40] Sun Yanguang,Xu Chunyan,Yang Jian,et al.Frequency-spatial entanglement learning for camouflaged object detection[C]//Computer Vision-ECCV 2024.Cham:Springer,2024:342-360.
[41] Zhang Chenxi,Zhang Qing,Wu Jiayun.Rethinking camouflaged object detection via foreground-background interactive learning[C]//ICASSP 2025-2025 IEEE International Conference on Acoustics,Speech and Signal Processing (ICASSP).IEEE,2025:1-5.
[42] Wang Tingran,Yu Zaiyang,Fang Jianwei,et al.Multidimensional fusion of frequency and spatial domain information for enhanced camouflaged object detection[J].Information Fusion,2025,117:102871.
[43] Xie Chen,Xia Changqun,Yu Tianshu,et al.Frequency representation integration for camouflaged object detection[C]//Proceedings of the 31st ACM International Conference on Multimedia.Ottawa,Canada:ACM,2023:1789-1797.
[44] Kirillov A,Mintun E,Ravi N,et al.Segment anything[C]//Proceedings of the IEEE/CVF ICCV.Paris,France,2023:4015-4026.
[45] Carion N,Massa F,Synnaeve G,et al.End-to-end object detection with transformers[C]//Proceedings of the European Conference on Computer Vision.Cham:Springer,2020:213-229.
[46] Yang Fan,Zhai Qiang,Li Xin,et al.Uncertainty-guided transformer reasoning for camouflaged object detection[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.Montreal,Canada:IEEE,2021:4146-4155.
[47] Gao Ruijun,Guo Qing,Xu Juefei,et al.Can you spot the chameleon? Adversarially camouflaging images from co-salient models[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans,USA:IEEE,2022:2150-2159.
[48] Han Tianxin,Wang Xingwei,Dong Qing,et al.Weakly supervised camouflaged object detection as progressive perception learning[J].Knowledge-Based Systems,2025,325:113993.
[49] Chen Huafeng,Shao Dian,Guo Guangqian,et al.Just a hint:point-supervised camouflaged object detection[C]//European Conference on Computer Vision.Cham:Springer Nature Switzerland,2024:332-348.
[50] Meinhardt T,Kirillov A,Leal-Taixé L,et al.Trackformer:multi-object tracking with transformers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville,USA,2021:8844-8854.
[51] Du Ji,Hao Fangwei,Yu Mingyang,et al.Shift the lens:environment-aware unsupervised camouflaged object detection[C]//Proceedings of the Computer Vision and Pattern Recognition Conference,2025:19271-19282.
[52] 彭锐晖,赖杰,孙殿星,等.一种改进YOLOv5算法的伪装目标检测方法[J].电子与信息学报,2024,46(8):3324-3333.
[53] Xu Xiuqi,Zhu Mingyu,Yu Jinhao,et al.Boundary guidance network for camouflage object detection[J].Image and Vision Computing,2021,114:134283.
[54] 程擎,何汶键,贾志杰,等.基于改进YOLOv7的低空小型无人机识别算法研究[J].贵州师范大学学报(自然科学版),2025,43(5):45-53.
[55] 张学凯,彭月平,唐威,等.隐蔽伪装目标检测方法综述[J].陆军工程大学学报,2025,4(1):61-70.
[56] Cai Wei,Gao Weijie,Ding Yao,et al.Diffusion model for camouflaged object segmentation with frequency domain[J].Electronics,2024,13(19):3922.
[57] Chen Zefeng,Li Zhijiang,Xue Yunqi,et al.Foundation model based camouflaged object detection[J].IET Computer Vision,2025,19(1):e70009.
[58] 韩烜宇,王安志,杨成帮,等.面向计算机视觉应用的扩散模型综述[J].贵州师范大学学报(自然科学版),2025,43(1):115-128.
[59] Zhai Qiang,Li Xin,Yang Fan,et al.Mutual graph learning for camouflaged object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition,2021:12997-13007.

基金资助

国家自然科学基金项目(62162013);贵州省科技厅基础研究计划项目(黔科合基础MS〔2025〕249)