低干信比SAR非虚假目标类有源干扰的轻量化鉴别网络

王重淞 蒲巍 高杰 毛德庆 刘欣远 武俊杰 黄钰林 杨建宇

王重淞, 蒲巍, 高杰, 等. 低干信比SAR非虚假目标类有源干扰的轻量化鉴别网络[J]. 雷达学报(中英文), 待出版. doi: 10.12000/JR25195
引用本文: 王重淞, 蒲巍, 高杰, 等. 低干信比SAR非虚假目标类有源干扰的轻量化鉴别网络[J]. 雷达学报(中英文), 待出版. doi: 10.12000/JR25195
WANG Chongsong, PU Wei, GAO Jie, et al. Lightweight discrimination network for non-spoofing active jamming in SAR under low JSR[J]. Journal of Radars, in press. doi: 10.12000/JR25195
Citation: WANG Chongsong, PU Wei, GAO Jie, et al. Lightweight discrimination network for non-spoofing active jamming in SAR under low JSR[J]. Journal of Radars, in press. doi: 10.12000/JR25195

低干信比SAR非虚假目标类有源干扰的轻量化鉴别网络

DOI: 10.12000/JR25195 CSTR: 32380.14.JR25195
基金项目: 国家自然科学基金(U25B2015)
详细信息
    作者简介:

    王重淞,博士生,主要研究方向为合成孔径雷达抗干扰和深度学习等

    蒲 巍,博士,教授,主要研究方向为合成孔径雷达成像、雷达信号处理、智能信号处理等

    高 杰,硕士生,主要研究方向为智能化合成孔径雷达干扰鉴别和抑制

    毛德庆,博士,副教授,主要研究方向为雷达探测成像、雷达超分辨成像、雷达智能化成像等

    刘欣远,博士生,主要研究方向为智能化雷达抗干扰等

    武俊杰,博士,教授,主要研究方向为双/多基SAR、智能化雷达成像等

    黄钰林,博士,教授,主要研究方向为雷达成像、检测与识别和机器学习等

    杨建宇,博士,教授,主要研究方向为前视雷达成像、双/多基SAR成像、新体制雷达探测与成像等

    通讯作者:

    蒲巍 puwei@uestc.edu.cn

    责任主编:孙光才 Corresponding Editor: SUN Guangcai

  • 中图分类号: TN958

Lightweight Discrimination Network for Non-spoofing Active Jamming in SAR under Low JSR

Funds: The National Natural Science Foundation of China (U25B2015)
More Information
  • 摘要: 合成孔径雷达(SAR)以其全天时、全天候及高分辨成像优势,在军事侦察与遥感监测等领域发挥着关键作用。然而,现代复杂电磁环境中的多样化干扰手段,严重破坏SAR回波信号特征,导致成像结果出现模糊、失真乃至目标完全不可辨识等问题。鉴于不同类型干扰在形成机理和抑制策略上的本质差异,精确的干扰鉴别成为实现有效抗干扰的核心前提。当前SAR干扰鉴别方法仍面临两大挑战:其一,在干扰信号与目标信号能量相近时,干扰特征易被目标能量掩盖,难以可靠检测与鉴别;其二,现有鉴别网络普遍复杂度过高、实时性差,难以满足实际工程应用需求。针对上述问题,该文提出一种基于轻量化网络的低干信比SAR非虚假目标类有源干扰鉴别网络。该方法创新性地引入了格型变换模块与超核感知模块。其中,格型变换模块旨在强化对干扰目标的细粒度特征提取能力,进而显著提升了低干信比条件下的干扰鉴别性能;超核感知模块则基于点目标成像特性,设计了超核卷积,该卷积在增强上下文信息捕获能力的同时,亦实现了算法的轻量化。实验部分通过多维度评估验证了方法的优越性,包括模块的有效性分析、不同模型的精度-复杂度权衡分析,以及不同干信比下的鲁棒性测试。结果表明,所提方法在低干信比条件下仍能保持较高鉴别性能,同时计算效率满足实时性需求。

     

  • 图  1  不同干扰功率的成像结果示意图

    Figure  1.  Illustration of imaging results under different interference powers

    图  2  SAR系统几何构型

    Figure  2.  SAR system geometry

    图  3  间歇采样转发类干扰的不同策略(a) 一个合成孔径回波;(b) 直接转发;(c) 重复转发;(d) 循环转发

    Figure  3.  Implementation approaches of sampled-repeater jamming (a) Synthetic aperture echo; (b) ISRJ; (c) ISRRJ; (d) ISCRJ

    图  4  不同干扰样式的干扰下成像结果

    Figure  4.  Imaging results under different jamming

    图  5  不同JSR下的压制干扰结果对比图

    Figure  5.  Comparative diagram of suppression interference results under different JSR conditions

    图  6  不同JSR下的转发类干扰结果对比图

    Figure  6.  Comparative diagram of forwarding class interference results under different JSR conditions

    图  7  轻量化有源干扰鉴别网络结构

    Figure  7.  Lightweight network architecture for active jamming recognition

    图  8  所提格型变换模块结构

    Figure  8.  The proposed structure of lattice transform block

    图  9  $ H(\cdot ) $ 的卷积结构

    Figure  9.  CNN block of $ H(\cdot ) $

    图  10  SAR点目标成像结果

    Figure  10.  SAR point target imaging

    图  11  超核卷积的结构

    Figure  11.  Structure of the Hyperkernel Block

    图  12  数据集仿真流程图

    Figure  12.  Flowchart of dataset simulation

    图  13  无干扰下的干扰数据集数据切片

    Figure  13.  Dataset samples without jamming

    图  14  模型训练过程损失曲线

    Figure  14.  Loss value of models in training process

    图  15  不同结构的热力图(Grad-CAM)对比结果(第1行展示了没有使用格型变换模块的可视化,第2行展示了带有格型变换模块的可视化)

    Figure  15.  Comparison of heatmap (Grad-CAM) results from different structures (the first row displays visualizations without the LT Block, while the second row presents corresponding visualizations incorporating the LT Block)

    图  16  所提方法鉴别结果的混淆矩阵

    Figure  16.  The confusion matrix of the discrimination results of the proposed method

    图  17  不同JSR下的模型性能

    Figure  17.  Model performance under different JSR conditions

    图  18  哨兵1号实测数据

    Figure  18.  Sentinel-1 measured data

    表  1  干扰样式编码

    Table  1.   Jamming coding

    干扰类型编号干扰类型编号
    无干扰C0方位向均匀间歇采样转发干扰C1
    方位向非均匀间歇采样转发干扰C2距离向均匀间歇采样转发干扰C3
    距离向非均匀间歇采样转发干扰C4二维均匀间歇采样转发干扰C5
    二维非均匀间歇采样转发干扰C6均匀间歇采样重复转发干扰C7
    非均匀间歇采样循环转发干扰C8乘性调制噪声干扰C9
    步进移频干扰C10分段移频干扰C11
    随机移频干扰C12失配干扰C13
    噪声压制干扰C14一维卷积噪声干扰C15
    二维卷积噪声干扰C16窄带射频干扰C17
    正弦调频宽带射频干扰C18线性调频宽带射频干扰C19
    下载: 导出CSV

    表  2  消融实验结果(%)

    Table  2.   Ablation study results (%)

    模型结构 OA AP AR F1
    基线模型 85.66 84.44 85.98 85.20
    基线模型+超核卷积模块 91.44 93.90 95.25 94.57
    基线模型+格型变换模块+
    超核卷积模块
    99.57 99.56 99.63 99.60
    下载: 导出CSV

    表  3  各种鉴别方法的准确率性能(%)

    Table  3.   Comparison of accuracy for various discrimination methods (%)

    编号 VGG-16 ResNet-18 Inception v4 Xception Mobilenet V2 VGG-16+AN Ours
    C0 99.12 98.69 96.78 94.96 93.06 99.42 100.00
    C1 100.00 93.89 100.00 93.45 96.33 99.13 100.00
    C2 91.27 95.63 90.39 84.72 96.33 100.00 100.00
    C3 96.07 92.14 98.69 84.28 94.50 99.13 99.13
    C4 90.39 74.24 59.83 74.22 80.73 98.25 98.25
    C5 96.94 90.83 96.07 73.36 86.24 94.76 100.00
    C6 100.00 96.07 98.25 99.13 95.41 97.82 100.00
    C7 97.82 60.26 96.94 94.76 69.72 99.13 99.13
    C8 82.10 78.17 86.03 76.86 74.31 95.20 100.00
    C9 96.51 95.20 99.56 95.63 93.58 96.07 98.25
    C10 82.53 86.03 82.53 79.91 85.32 84.72 96.51
    C11 100.00 100.00 99.56 100.00 100.00 100.00 100.00
    C12 94.32 90.39 93.45 88.21 94.50 98.25 100.00
    C13 100.00 100.00 100.00 98.69 100.00 100.00 100.00
    C14 100.00 99.56 89.08 99.56 99.08 100.00 100.00
    C15 100.00 95.63 100.0 98.25 96.33 100.00 100.00
    C16 100.00 99.13 95.20 82.53 98.17 99.56 100.00
    C17 100.00 99.13 100.00 100.00 98.17 100.00 100.00
    C18 100.00 98.25 99.56 98.69 98.17 100.00 100.00
    C19 100.00 99.56 97.82 95.63 99.08 100.00 100.00
    注:加粗数值表示最优。
    下载: 导出CSV

    表  4  各种鉴别方法的性能指标

    Table  4.   Performance metrics of various discrimination models

    指标 VGG-16 ResNet-18 Inception v4 Xception Mobilenet V2 VGG-16+AN Ours
    OA(%) 96.42 92.35 94.05 90.76 92.51 98.10 99.57
    AP(%) 96.35 92.14 93.99 90.64 92.56 98.07 99.56
    AR(%) 96.84 92.48 94.92 91.79 92.45 99.42 99.63
    F1 (%) 96.66 92.31 93.99 91.21 92.51 98.24 99.60
    Kappa (%) 96.23 91.94 93.74 90.26 92.04 98.00 99.55
    Flops(G) 80.7 9.5 37.8 2.4 1.7 365 3.5
    Params(M) 134 11.19 41.17 20.85 2.25 2338.06 4.32
    TPI(s) 0.51 0.09 0.47 0.42 0.15 3.44 0.21
    注:加粗数值表示最优。
    下载: 导出CSV
  • [1] SADIQ R, QURESHI M B, and KHAN M M. De-convolution and De-noising of SAR based GPS images using hybrid particle swarm optimization[J]. Chinese Journal of Electronics, 2023, 32(1): 166–176. doi: 10.23919/cje.2021.00.138.
    [2] ZHOU Lifan, ZHOU Xuanyu, FENG Huanghao, et al. Transformer-based semantic segmentation for flood region recognition in SAR images[J]. IEEE Journal on Miniaturization for Air and Space Systems, 2025, 6(3): 222–229. doi: 10.1109/JMASS.2025.3542124.
    [3] LIU Niantang, ZHAO Qunshan, WILLIAMS R, et al. Enhanced crop mapping using polarimetric SAR features and time series deep learning: A case study in Bei’an, China[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5002917. doi: 10.1109/TGRS.2025.3544339.
    [4] PEREIRA-PIRES J, GUERRA-HERNÁNDEZ J, SILVA J M N, et al. Forest height mapping with multifrequency SAR in mediterranean forests[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 4403415. doi: 10.1109/TGRS.2025.3538216.
    [5] 武俊杰, 杨建宇, 李中余, 等. 双基地SAR成像处理方法综述[J]. 雷达学报(中英文), 2025, 14(5): 1115–1141. doi: 10.12000/JR25067.

    WU Junjie, YANG Jianyu, LI Zhongyu, et al. Review of bistatic synthetic aperture radar imaging methods[J]. Journal of Radars, 2025, 14(5): 1115–1141. doi: 10.12000/JR25067.
    [6] CAO Rui, WANG Yong, GIUSTI E, et al. 3-D reconstruction of ship target based on SAR images sequence and scatterer tracking technique[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5200415. doi: 10.1109/TGRS.2024.3514699.
    [7] WU Wanmin, PU Wei, HAI Yu, et al. A deep learning-based SAR imaging framework for ship targets with sample-wise variant motion[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5214116. doi: 10.1109/TGRS.2025.3575325.
    [8] DOMÍNGUEZ E M, BROTZER P, CASALINI E, et al. Mapping urban areas and infrastructure through fusion of airborne SAR 3-D images: A comparative study with ALS sensors[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025, 18: 6164–6181. doi: 10.1109/JSTARS.2025.3541425.
    [9] WANG Chongsong, LI Yuzhao, SHANG Yuanzhe, et al. An anchor-free method for aircraft detection in SAR images based on density map[C]. 2024 IEEE International Conference on Signal, Information and Data Processing, Zhuhai, China, 2024: 1–6. doi: 10.1109/ICSIDP62679.2024.10868269.
    [10] 黄岩, 赵博, 陶明亮, 等. 合成孔径雷达抗干扰技术综述[J]. 雷达学报, 2020, 9(1): 86–106. doi: 10.12000/JR19113.

    HUANG Yan, ZHAO Bo, TAO Mingliang, et al. Review of synthetic aperture radar interference suppression[J]. Journal of Radars, 2020, 9(1): 86–106. doi: 10.12000/JR19113.
    [11] 李本朋. 国外机载箔条干扰技术的发展[J]. 机械管理开发, 2018, 33(2): 56–58. doi: 10.16525/j.cnki.cn14-1134/th.2018.02.23.

    LI Benpeng. Development of foreign airborne chaff jamming technology[J]. Mechanical Management and Development, 2018, 33(2): 56–58. doi: 10.16525/j.cnki.cn14-1134/th.2018.02.23.
    [12] 李超, 李芳. 基于人工电磁材料的新型电磁隐身机制——电磁隐身斗篷[J]. 北京石油化工学院学报, 2009, 17(1): 48–52. doi: 10.3969/j.issn.1008-2565.2009.01.011.

    LI Chao and LI Fang. A novel electromagnetic stealth method-electromagnetic invisible cloaks[J]. Journal of Beijing Institute of Petrochemical Technology, 2009, 17(1): 48–52. doi: 10.3969/j.issn.1008-2565.2009.01.011.
    [13] LIN Hao, XING Mengdao, LOU Yishan, et al. Research on anti-deception forwarding interference of squint azimuth multichannel SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5203912. doi: 10.1109/TGRS.2023.3334022.
    [14] WANG Zan, GUO Zhengwei, SHU Gaofeng, et al. Radar jamming recognition: Models, methods, and prospects[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025, 18: 3315–3343. doi: 10.1109/JSTARS.2024.3522951.
    [15] 朱吉利, 范振军. 一种基于集成学习的干扰频点检测方法[J]. 图像与信号处理, 2025, 14(1): 100–107. doi: 10.12677/jisp.2025.141010.

    ZHU Jili and FAN Zhenjun. A detection method of jamming frequency based on ensemble learning[J]. Journal of Image and Signal Processing, 2025, 14(1): 100–107. doi: 10.12677/jisp.2025.141010.
    [16] LV Qinzhe, QUAN Yinghui, FENG Wei, et al. Radar deception jamming recognition based on weighted ensemble CNN with transfer learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5107511. doi: 10.1109/TGRS.2021.3129645.
    [17] QU Qizhe, WEI Shujun, LIU Shan, et al. JRNet: Jamming recognition networks for radar compound suppression jamming signals[J]. IEEE Transactions on Vehicular Technology, 2020, 69(12): 15035–15045. doi: 10.1109/TVT.2020.3032197.
    [18] LUO Zhenyu, CAO Yunhe, YEO T S, et al. Few-shot radar jamming recognition network via time-frequency self-attention and global knowledge distillation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 5105612. doi: 10.1109/TGRS.2023.3280322.
    [19] 邢世其, 纪朋徽, 代大海, 等. 方位向调制干扰对高分宽幅多通道SAR的影响[J]. 系统工程与电子技术, 2024, 46(6): 1946–1956. doi: 10.12305/j.issn.1001-506X.2024.06.12.

    XING Shiqi, JI Penghui, DAI Dahai, et al. Influence of azimuth-modulation jamming on high-resolution wide-swath multi-channel SAR[J]. Systems Engineering and Electronics, 2024, 46(6): 1946–1956. doi: 10.12305/j.issn.1001-506X.2024.06.12.
    [20] 康晓磊. 面向目标识别的SAR成像干扰类型判别与分级方法研究[D]. [硕士论文], 华中科技大学, 2018. doi: 10.7666/d.D01544285.

    KANG Xiaolei. Research on discrimination and classification methods of SAR jamming types for target recognition[D]. [Master dissertation], Huazhong University of Science and Technology, 2018. doi: 10.7666/d.D01544285.
    [21] 陈思伟, 崔兴超, 李铭典, 等. 基于深度CNN模型的SAR图像有源干扰类型识别方法[J]. 雷达学报, 2022, 11(5): 897–908. doi: 10.12000/JR22143.

    CHEN Siwei, CUI Xingchao, LI Mingdian, et al. SAR image active jamming type recognition based on deep CNN model[J]. Journal of Radars, 2022, 11(5): 897–908. doi: 10.12000/JR22143.
    [22] 汪日超. 星载SAR压制式干扰与抗干扰技术研究[D]. [硕士论文], 电子科技大学, 2017.

    WANG Richao. Research on spaceborne SAR compression interference and anti - jamming technology[D]. [Master dissertation], University of Electronic Science and Technology of China, 2017.
    [23] 黄大通, 邢世其, 李永祯, 等. 基于乘积调制的SAR灵巧干扰方法[J]. 系统工程与电子技术, 2021, 43(11): 3160–3168. doi: 10.12305/j.issn.1001-506X.2021.11.15.

    HUANG Datong, XING Shiqi, LI Yongzhen, et al. Smart jamming method against SAR based on multiplication modulation[J]. Systems Engineering and Electronics, 2021, 43(11): 3160–3168. doi: 10.12305/j.issn.1001-506X.2021.11.15.
    [24] 房明星, 王杰贵, 雷磊. SAR雷达二维噪声卷积调制干扰研究[J]. 现代防御技术, 2014, 42(2): 139–144, 160. doi: 10.3969/j.issn.1009-086x.2014.02.025.

    FANG Mingxing, WANG Jiegui, and LEI Lei. Study on 2D noise convolution modulation jamming to SAR[J]. Modern Defence Technology, 2014, 42(2): 139–144, 160. doi: 10.3969/j.issn.1009-086x.2014.02.025.
    [25] 丁毅. 基于多维域分析的SAR射频干扰抑制方法研究[D]. [博士论文], 西安电子科技大学, 2022. doi: 10.27389/d.cnki.gxadu.2022.003158.

    DING Yi. Research on SAR radio frequency interference mitigation method using multidimensional analysis[D]. [Ph.D. dissertation], Xidian University, 2022. doi: 10.27389/d.cnki.gxadu.2022.003158.
    [26] SHEN Jiayuan, HAN Bing, PAN Zongxu, et al. Learning time-frequency information with prior for SAR radio frequency interference suppression[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5239716. doi: 10.1109/TGRS.2022.3225499.
    [27] LI Ning, ZHANG Hengrui, LV Zongsen, et al. Simultaneous screening and detection of RFI from massive SAR images: A case study on European sentinel-1[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5231917. doi: 10.1109/TGRS.2022.3191815.
    [28] 刘一兵, 罗强, 刘记红, 等. 基于分段移频调制的间歇采样重复转发干扰[J]. 电子信息对抗技术, 2023, 38(4): 5–12. doi: 10.3969/j.issn.1674-2230.2023.04.002.

    LIU Yibing, LUO Qiang, LIU Jihong, et al. Interrupted-sampling repetitive repeater jamming based on segmented shift-frequency modulation[J]. Electronic Information Warfare Technology, 2023, 38(4): 5–12. doi: 10.3969/j.issn.1674-2230.2023.04.002.
    [29] 张养瑞, 李云杰, 李曼玲, 等. 间歇采样非均匀重复转发实现多假目标压制干扰[J]. 电子学报, 2016, 44(1): 46–53. doi: 10.3969/j.issn.0372-2112.2016.01.008.

    ZHANG Yangrui, LI Yunjie, LI Manling, et al. Suppress jamming technique of multiple false targets on interrupted-sampling and non-uniform periodic repeater[J]. Acta Electronica Sinica, 2016, 44(1): 46–53. doi: 10.3969/j.issn.0372-2112.2016.01.008.
    [30] 孙宗正, 刘智星, 肖国尧, 等. 非均匀间歇采样转发干扰对脉内捷变雷达影响分析[J]. 系统工程与电子技术, 2024, 46(5): 1544–1554. doi: 10.12305/j.issn.1001-506X.2024.05.09.

    SUN Zongzheng, LIU Zhixing, XIAO Guoyao, et al. Analysis of the influence of non uniform interrupted sampling repeater jamming on intra-pulse agile radar[J]. Systems Engineering and Electronics, 2024, 46(5): 1544–1554. doi: 10.12305/j.issn.1001-506X.2024.05.09.
    [31] 周政, 唐宏, 张永顺. LFM脉压雷达的随机移频干扰研究[J]. 现代防御技术, 2010, 38(1): 103–108. doi: 10.3969/j.issn.1009-086x.2010.01.023.

    ZHOU Zheng, TANG Hong, and ZHANG Yongshun. Randomly shift frequency jamming to LFM pulse compression radar[J]. Modern Defence Technology, 2010, 38(1): 103–108. doi: 10.3969/j.issn.1009-086x.2010.01.023.
    [32] 黄洪旭, 黄知涛, 周一宇. 对合成孔径雷达的随机移频干扰[J]. 信号处理, 2007, 23(1): 41–45. doi: 10.3969/j.issn.1003-0530.2007.01.009.

    HUANG Hongxu, HUANG Zhitao, and ZHOU Yiyu. Randomly-shift-frequency jamming style to SAR[J]. Signal Processing, 2007, 23(1): 41–45. doi: 10.3969/j.issn.1003-0530.2007.01.009.
    [33] 蔡幸福, 张雄美, 宋建社, 等. 基于脉间分段随机移频的合成孔径雷达干扰技术及其应用模型[J]. 兵工学报, 2015, 36(11): 2196–2202. doi: 10.3969/j.issn.1000-1093.2015.11.027.

    CAI Xingfu, ZHANG Xiongmei, SONG Jianshe, et al. A jamming approach to SAR based on inter-pulse subsection random frequency-shift technique and its application[J]. Acta Armamentarii, 2015, 36(11): 2196–2202. doi: 10.3969/j.issn.1000-1093.2015.11.027.
    [34] 和小冬, 李昀豪, 祝俊, 等. 合成孔径雷达二维失配压制干扰方法[J]. 电子信息对抗技术, 2014, 29(3): 24–28, 79. doi: 10.3969/j.issn.1674-2230.2014.03.006.

    HE Xiaodong, LI Yunhao, ZHU Jun, et al. A Bi-dimensional mismatching suppressed jamming for countering SAR[J]. Electronic Information Warfare Technology, 2014, 29(3): 24–28, 79. doi: 10.3969/j.issn.1674-2230.2014.03.006.
    [35] LI Bodong and GAO Xieping. Lattice structure for regular linear phase paraunitary filter bank with odd decimation factor[J]. IEEE Signal Processing Letters, 2014, 21(1): 14–17. doi: 10.1109/LSP.2013.2285435.
    [36] LUO Xiaotong, XIE Yuan, ZHANG Yulun, et al. LatticeNet: Towards lightweight image super-resolution with lattice block[C]. 16th European Conference on Computer Vision, Glasgow, UK, 2020. doi: 10.1007/978-3-030-58542-6_17.
    [37] WANG Zheyuan, LI Liangliang, XUE Yuan, et al. FeNet: Feature enhancement network for lightweight remote-sensing image super-resolution[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5622112. doi: 10.1109/TGRS.2022.3168787.
    [38] DING Xiaohan, ZHANG Yiyuan, GE Yixiao, et al. UniRepLKNet: A universal perception large-kernel ConvNet for audio, video, point cloud, time-series and image recognition[C]. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, USA, 2024: 5513–5524. doi: 10.1109/CVPR52733.2024.00527.
    [39] CHEN Honghao, CHU Xiangxiang, REN Yongjian, et al. PeLK: Parameter-efficient large kernel ConvNets with peripheral convolution[C]. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, USA, 2024: 5557–5567. doi: 10.1109/CVPR52733.2024.00531.
    [40] HU Jie, SHEN Li, and SUN Gang. Squeeze-and-excitation networks[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 7132–7141. doi: 10.1109/CVPR.2018.00745.
    [41] MAO Anqi, MOHRI M, and ZHONG Yutao. Cross-entropy loss functions: Theoretical analysis and applications[C]. The 40th International Conference on Machine Learning, Honolulu, USA, 2023.
    [42] SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: Visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision, 2020, 128(2): 336–359. doi: 10.1007/s11263-019-01228-7.
    [43] SIMONYAN K and ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409. 1556, 2014.
    [44] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770–778. doi: 10.1109/CVPR.2016.90.
    [45] SZEGEDY C, IOFFE S, VANHOUCKE V, et al. Inception-v4, inception-ResNet and the impact of residual connections on learning[C]. The Thirty-First AAAI Conference on Artificial Intelligence, San Francisco, USA, 2016: 4278–4284.
    [46] CHOLLET F. Xception: Deep learning with depthwise separable convolutions[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 1800–1807. doi: 10.1109/CVPR.2017.195.
    [47] SANDLER M, HOWARD A, ZHU Menglong, et al. MobileNetV2: Inverted residuals and linear bottlenecks[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 4510–4520. doi: 10.1109/CVPR.2018.00474.
  • 加载中
图(18) / 表(4)
计量
  • 文章访问数: 
  • HTML全文浏览量: 
  • PDF下载量: 
  • 被引次数: 0
出版历程
  • 收稿日期:  2025-09-29
  • 修回日期:  2025-12-10
  • 网络出版日期:  2025-12-29

目录

    /

    返回文章
    返回