Acta Armamentarii ›› 2023, Vol. 44 ›› Issue (10): 3165-3176.doi: 10.12382/bgxb.2022.0605
Previous Articles Next Articles
WANG Qiang1,2,3, WU Letian1,2, LI Hong1,2, WANG Yong3, WANG Huan4, YANG Wankou1,2,*()
Received:
2022-07-05
Online:
2023-10-30
Contact:
YANG Wankou
CLC Number:
WANG Qiang, WU Letian, LI Hong, WANG Yong, WANG Huan, YANG Wankou. An Infrared Small Target Detection Method via Dual Network Collaboration[J]. Acta Armamentarii, 2023, 44(10): 3165-3176.
Add to citation manager EndNote|Ris|BibTeX
数据集 | 算法 | Precision↑ | Recall↑ | F1-measure↑ |
---|---|---|---|---|
FCN-RSTN[ | 0.12 | 0.75 | 0.21 | |
cGAN[ | 0.13 | 0.60 | 0.21 | |
All Seqs | U-net[ | 0.01 | 0.01 | 0.01 |
DeepLabv3[ | 0.17 | 0.59 | 0.27 | |
DualNet | 0.36 | 0.44 | 0.40 | |
FCN-RSTN[ | 0.54 | 0.39 | 0.45 | |
cGAN[ | 0.28 | 0.71 | 0.40 | |
Single | U-net[ | 0.01 | 0.02 | 0.01 |
DeepLabv3[ | 0.22 | 0.57 | 0.31 | |
DualNet | 0.52 | 0.54 | 0.53 |
Table 1 Experimental results using different methods
数据集 | 算法 | Precision↑ | Recall↑ | F1-measure↑ |
---|---|---|---|---|
FCN-RSTN[ | 0.12 | 0.75 | 0.21 | |
cGAN[ | 0.13 | 0.60 | 0.21 | |
All Seqs | U-net[ | 0.01 | 0.01 | 0.01 |
DeepLabv3[ | 0.17 | 0.59 | 0.27 | |
DualNet | 0.36 | 0.44 | 0.40 | |
FCN-RSTN[ | 0.54 | 0.39 | 0.45 | |
cGAN[ | 0.28 | 0.71 | 0.40 | |
Single | U-net[ | 0.01 | 0.02 | 0.01 |
DeepLabv3[ | 0.22 | 0.57 | 0.31 | |
DualNet | 0.52 | 0.54 | 0.53 |
模型 | 数据集 | Precision↑ | Recall↑ | F1-measure↑ |
---|---|---|---|---|
Dilated-Conv | All Seqs | 0.32 | 0.36 | 0.34 |
Single | 0.45 | 0.44 | 0.45 | |
ASPP | All Seqs | 0.34 | 0.42 | 0.38 |
Single | 0.48 | 0.46 | 0.47 | |
ASPP+PPM | All Seqs | 0.32 | 0.38 | 0.35 |
Single | 0.46 | 0.45 | 0.45 | |
ASPP+SPP | All Seqs | 0.34 | 0.48 | 0.40 |
Single | 0.49 | 0.53 | 0.51 | |
ASPP+SPP+ | All Seqs | 0.46 | 0.44 | 0.40 |
scSE | Single | 0.52 | 0.53 | 0.52 |
ASPP+SPP+ | All Seqs | 0.36 | 0.44 | 0.40 |
scSE+DCNv2 | Single | 0.52 | 0.54 | 0.53 |
Table 3 Ablation experiments of DualNet method
模型 | 数据集 | Precision↑ | Recall↑ | F1-measure↑ |
---|---|---|---|---|
Dilated-Conv | All Seqs | 0.32 | 0.36 | 0.34 |
Single | 0.45 | 0.44 | 0.45 | |
ASPP | All Seqs | 0.34 | 0.42 | 0.38 |
Single | 0.48 | 0.46 | 0.47 | |
ASPP+PPM | All Seqs | 0.32 | 0.38 | 0.35 |
Single | 0.46 | 0.45 | 0.45 | |
ASPP+SPP | All Seqs | 0.34 | 0.48 | 0.40 |
Single | 0.49 | 0.53 | 0.51 | |
ASPP+SPP+ | All Seqs | 0.46 | 0.44 | 0.40 |
scSE | Single | 0.52 | 0.53 | 0.52 |
ASPP+SPP+ | All Seqs | 0.36 | 0.44 | 0.40 |
scSE+DCNv2 | Single | 0.52 | 0.54 | 0.53 |
数据集 | α | β | F1-measure↑ |
---|---|---|---|
0.05 | 0.95 | 0.38 | |
0.1 | 0.9 | 0.40 | |
0.2 | 0.8 | 0.36 | |
0.3 | 0.7 | 0.37 | |
All Seqs | 0.4 | 0.6 | 0.34 |
0.5 | 0.5 | 0.33 | |
0.6 | 0.4 | 0.31 | |
0.7 | 0.3 | 0.31 | |
0.8 | 0.2 | 0.30 | |
0.05 | 0.95 | 0.51 | |
0.1 | 0.9 | 0.53 | |
0.2 | 0.8 | 0.51 | |
0.3 | 0.7 | 0.51 | |
Single | 0.4 | 0.6 | 0.50 |
0.5 | 0.5 | 0.50 | |
0.6 | 0.4 | 0.49 | |
0.7 | 0.3 | 0.49 | |
0.8 | 0.2 | 0.47 |
Table 4 Influence of α and β on DualNet method
数据集 | α | β | F1-measure↑ |
---|---|---|---|
0.05 | 0.95 | 0.38 | |
0.1 | 0.9 | 0.40 | |
0.2 | 0.8 | 0.36 | |
0.3 | 0.7 | 0.37 | |
All Seqs | 0.4 | 0.6 | 0.34 |
0.5 | 0.5 | 0.33 | |
0.6 | 0.4 | 0.31 | |
0.7 | 0.3 | 0.31 | |
0.8 | 0.2 | 0.30 | |
0.05 | 0.95 | 0.51 | |
0.1 | 0.9 | 0.53 | |
0.2 | 0.8 | 0.51 | |
0.3 | 0.7 | 0.51 | |
Single | 0.4 | 0.6 | 0.50 |
0.5 | 0.5 | 0.50 | |
0.6 | 0.4 | 0.49 | |
0.7 | 0.3 | 0.49 | |
0.8 | 0.2 | 0.47 |
损失函数 | 数据集 | F1-measure↑ |
---|---|---|
Focal Loss | All Seqs | 0 |
Single | 0 | |
L1 Loss | All Seqs | 0 |
Single | 0 | |
Balanced L1 Loss | All Seqs | 0 |
Single | 0 | |
Smooth L1 Loss | All Seqs | 0.40 |
Single | 0.53 | |
L2 Loss | All Seqs | 0.34 |
Single | 0.47 |
Table 5 Influence of different loss functions on DualNet method
损失函数 | 数据集 | F1-measure↑ |
---|---|---|
Focal Loss | All Seqs | 0 |
Single | 0 | |
L1 Loss | All Seqs | 0 |
Single | 0 | |
Balanced L1 Loss | All Seqs | 0 |
Single | 0 | |
Smooth L1 Loss | All Seqs | 0.40 |
Single | 0.53 | |
L2 Loss | All Seqs | 0.34 |
Single | 0.47 |
数据集 | λ1 | λ2 | F1-measure↑ |
---|---|---|---|
100 | 10 | 0.53 | |
All Seqs | 50 | 5 | 0.46 |
10 | 1 | 0.43 | |
100 | 10 | 0.40 | |
Single | 50 | 5 | 0.38 |
10 | 1 | 0.37 |
Table 6 Influence of different values of λ1 and λ2
数据集 | λ1 | λ2 | F1-measure↑ |
---|---|---|---|
100 | 10 | 0.53 | |
All Seqs | 50 | 5 | 0.46 |
10 | 1 | 0.43 | |
100 | 10 | 0.40 | |
Single | 50 | 5 | 0.38 |
10 | 1 | 0.37 |
模型 | 参数量/106 | 每张检测时间/ms |
---|---|---|
网络分支1 | 0.71 | 1 |
网络分支2 | 0.86 | 8 |
Table 7 Speed and model size of DualNet model
模型 | 参数量/106 | 每张检测时间/ms |
---|---|---|
网络分支1 | 0.71 | 1 |
网络分支2 | 0.86 | 8 |
方法 | IoU/10-2 | Pd/10-2 | Fa/10-2 |
---|---|---|---|
Top-Hat[ | 7.143 | 79.84 | 0.1000 |
Max-Median[ | 4.172 | 69.20 | 0.0060 |
WSLCM[ | 1.158 | 77.95 | 0.5400 |
TLLCM[ | 1.029 | 79.09 | 0.5899 |
IPI[ | 25.67 | 85.55 | 0.0011 |
NRAM[ | 12.16 | 74.52 | 0.0014 |
RIPT[ | 11.05 | 79.08 | 0.0023 |
PSTNN[ | 22.40 | 77.95 | 0.0029 |
MSLSTIPT[ | 10.30 | 82.13 | 0.1131 |
ACM[ | 70.33 | 93.91 | 0.0004 |
MDvsFA-cGAN[ | 60.30 | 89.35 | 0.0056 |
DualNet | 71.37 | 100.00 | 0.0014 |
Table 8 Generalization capability tests over SIRST dataset
方法 | IoU/10-2 | Pd/10-2 | Fa/10-2 |
---|---|---|---|
Top-Hat[ | 7.143 | 79.84 | 0.1000 |
Max-Median[ | 4.172 | 69.20 | 0.0060 |
WSLCM[ | 1.158 | 77.95 | 0.5400 |
TLLCM[ | 1.029 | 79.09 | 0.5899 |
IPI[ | 25.67 | 85.55 | 0.0011 |
NRAM[ | 12.16 | 74.52 | 0.0014 |
RIPT[ | 11.05 | 79.08 | 0.0023 |
PSTNN[ | 22.40 | 77.95 | 0.0029 |
MSLSTIPT[ | 10.30 | 82.13 | 0.1131 |
ACM[ | 70.33 | 93.91 | 0.0004 |
MDvsFA-cGAN[ | 60.30 | 89.35 | 0.0056 |
DualNet | 71.37 | 100.00 | 0.0014 |
[1] |
|
[2] |
|
[3] |
赵耀. 红外弱小目标检测算法研究[D]. 杭州: 浙江大学, 2015.
|
|
|
[4] |
万丽丽. 基于背景预测的红外目标检测[D]. 武汉: 华中科技大学, 2015.
|
|
|
[5] |
韩金辉, 董兴浩, 李知静, 等. 基于局部对比度机制的红外弱小目标检测算法[J]. 红外技术, 2021, 43(4): 357-366.
|
|
|
[6] |
危水根, 王程伟, 陈震, 等. 基于视觉注意机制的红外弱小目标检测[J]. 光子学报, 2021, 50(1): 0110001.
|
|
|
[7] |
朱硕雅, 杨德振, 贾鹏, 等. 时空联合红外小目标检测算法的设计与实现[J]. 红外与激光, 2021, 51(3): 388-392.
|
|
|
[8] |
娄康. 基于深度学习与生物视觉机制的红外弱小目标检测与跟踪方法研究[D]. 镇江: 江苏科技大学, 2020.
|
|
|
[9] |
黄乐弘. 基于深度学习的空中红外目标检测方法研究[D]. 北京: 中国科学院大学, 2020.
|
|
|
[10] |
|
[11] |
杨其利, 周炳红, 郑伟. 基于全卷积递归网络的弱小目标检测方法[J]. 光学学报, 2020, 40(13): 43-55.
|
|
|
[12] |
梁杰, 李磊, 任君. 基于深度学习的红外图像遮挡干扰检测方法[J]. 兵工学报, 2019, 40(7): 1401-1410.
doi: 10.3969/j.issn.1000-1093.2019.07.009 |
doi: 10.3969/j.issn.1000-1093.2019.07.009 |
|
[13] |
田萱, 王亮, 丁琪. 基于深度学习的图像语义分割方法综述[J]. 软件学报, 2019, 30(2): 440-468.
|
|
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
doi: 10.1109/TPAMI.2015.2389824 pmid: 26353135 |
[22] |
|
[23] |
|
[24] |
|
[25] |
doi: 10.1109/TPAMI.34 URL |
[26] |
|
[27] |
|
[28] |
|
[29] |
|
[30] |
|
[31] |
|
[32] |
|
[33] |
doi: 10.1109/TPAMI.34 URL |
[34] |
doi: 10.1117/1.600620 URL |
[35] |
|
[36] |
doi: 10.1109/LGRS.2020.3004978 URL |
[37] |
doi: 10.1109/LGRS.8859 URL |
[38] |
pmid: 24043387 |
[39] |
doi: 10.3390/rs10111821 URL |
[40] |
doi: 10.1109/JSTARS.4609443 URL |
[41] |
doi: 10.3390/rs11040382 URL |
[42] |
doi: 10.1109/TGRS.2020.3022069 URL |
[43] |
|
[44] |
|
[1] | TIAN Daming, MIAO Pu. Visible Light Communication Nonlinear Equalizer Based on Model Solving and Deep Learning [J]. Acta Armamentarii, 2024, 45(2): 466-473. |
[2] | ZHANG Kun, DU Ruiyi, SHI Haotian, HUA Shuai. Prediction of Aircraft Trajectory Based on Mogrifier-BiGRU [J]. Acta Armamentarii, 2024, 45(2): 373-384. |
[3] | YANG Jiaming, PAN Yue, WANG Qiang, CAO Huaigang, GAO Sunpei. Research on Deep Learning Method of Underwater Weak Target Tracking [J]. Acta Armamentarii, 2024, 45(2): 385-394. |
[4] | QIN Haolin, XU Tingfa, LI Jianan. Semi-supervised Hyperspectral Salient Object Detection Using Superpixel Attention and Siamese Structure [J]. Acta Armamentarii, 2023, 44(9): 2639-2649. |
[5] | HUI Kanghua, YANG Wei, LIU Haohan, ZHANG Zhi, ZHENG Jin, BAI Xiao. Enhanced Multi-scale Target Detection Method Based on YOLOv5 [J]. Acta Armamentarii, 2023, 44(9): 2600-2610. |
[6] | PENG Peiran, REN Shubo, LI Jianan, ZHOU Hongwei, XU Tingfa. Illumination-aware Multispectral Fusion Network for Pedestrian Detection [J]. Acta Armamentarii, 2023, 44(9): 2622-2630. |
[7] | ZHOU Yu, CAO Ronggang, LI Ping, MA Xiao. A Fuze Burst Point Detection Method for Outfield Test Images [J]. Acta Armamentarii, 2023, 44(8): 2453-2464. |
[8] | WANG Xuemin, YU Hongbo, ZHANG Xiangyu, AN Shu, LI Wenhai. Underwater Multi-target Detection Method Based on Hough Transform Track-before-detect Technique [J]. Acta Armamentarii, 2023, 44(7): 2114-2121. |
[9] | LIU Song, YAO Zhixiang, LU Daiqiang, YUAN Jun. An Automatic Detection Algorithm Based on Wideband Spatial Spectrum of Passive Sonar [J]. Acta Armamentarii, 2023, 44(6): 1764-1774. |
[10] | MA Pengge, WEI Hongguang, SUN Junling, TAO Ran, PANG Dongdong, SHAN Tao, CAI Zhiyong, LIU Zhaoyu. A LOG Filter Based Enhanced Local Contrast Algorithm to Detect Infrared Small Targets [J]. Acta Armamentarii, 2023, 44(4): 1041-1049. |
[11] | SUN Qiang, ZHANG Jiawei, YU Peng. Analysis of Electric Field Interference Characteristics and Signal Detection Method Based on Ocean Buoys [J]. Acta Armamentarii, 2023, 44(3): 857-864. |
[12] | WANG Yang, FENG Yongxin, SONG Bixue, TIAN Binghe. A Modulation Recognition Algorithm of DP-DRCnet Convolutional Neural Network [J]. Acta Armamentarii, 2023, 44(2): 545-555. |
[13] | ZHANG Liang'an, CHEN Yang, XIE Shenglong, LIU Tongxin. Crack Detection System for Aircraft Protective Grill based on Machine Vision and Deep Learning [J]. Acta Armamentarii, 2023, 44(2): 507-516. |
[14] | ZHENG Zhiwei, GUAN Xueyuan, FU Jian, MA Xunqiong, YIN Shang. Projectile Trajectory Prediction Based on CNN-LSTM Model [J]. Acta Armamentarii, 2023, 44(10): 2975-2983. |
[15] | WANG Zhong-mei, YANG Xiao-mei , GU Xing-fa. Small Target Detection in a Single Infrared Image Based on RPCA [J]. Acta Armamentarii, 2016, 37(9): 1753-1760. |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||