TY - GEN
T1 - Deep Learning Enabled Hidden Target Detection in Terahertz Images
AU - Tian, Nuoman
AU - Wang, Xingyu
AU - Xu, Yafei
AU - Wang, Rong
AU - Lian, Guanghui
AU - Zhang, Liuyang
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Because there is no ionizing radiation, terahertz wave has great application prospects in human security inspection. However, it is difficult to recognize hidden targets from terahertz images. At present, deep neural network is widely used in images recognition, especially gives great help for analyzing hidden targets in terahertz images from the noise interference. Here we have proposed an improved model based on YOLOX to improve the accuracy of existing neural network. To reduce the noise influence in THz images, the modified YOLOX network includes the addition of CBAM and ASFF module, the loss function of Loss(EIoU)or Loss(a-CloU)of the recognition task, and the loss function of Varifocal Loss of the positioning task. Our results can provide an insightful guideline to improve the hidden target detection and will be beneficial to the development of THz NDT technique.
AB - Because there is no ionizing radiation, terahertz wave has great application prospects in human security inspection. However, it is difficult to recognize hidden targets from terahertz images. At present, deep neural network is widely used in images recognition, especially gives great help for analyzing hidden targets in terahertz images from the noise interference. Here we have proposed an improved model based on YOLOX to improve the accuracy of existing neural network. To reduce the noise influence in THz images, the modified YOLOX network includes the addition of CBAM and ASFF module, the loss function of Loss(EIoU)or Loss(a-CloU)of the recognition task, and the loss function of Varifocal Loss of the positioning task. Our results can provide an insightful guideline to improve the hidden target detection and will be beneficial to the development of THz NDT technique.
KW - Nondestructive testing
KW - Target recognition
KW - Terahertz image
KW - YOLOX
UR - https://www.scopus.com/pages/publications/85150468593
U2 - 10.1109/ICSMD57530.2022.10058229
DO - 10.1109/ICSMD57530.2022.10058229
M3 - 会议稿件
AN - SCOPUS:85150468593
T3 - 2022 International Conference on Sensing, Measurement and Data Analytics in the Era of Artificial Intelligence, ICSMD 2022 - Proceedings
BT - 2022 International Conference on Sensing, Measurement and Data Analytics in the Era of Artificial Intelligence, ICSMD 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 3rd International Conference on Sensing, Measurement and Data Analytics in the Era of Artificial Intelligence, ICSMD 2022
Y2 - 22 December 2022 through 24 December 2022
ER -