Xiao Zhou
ABSTRACT
Because of small number of occupied pixels, lacking shape and texture information, the reliability of infrared remote target detection has always been a difficult research topic. To improve the accuracy and precision of detection of infrared small targets under complex background conditions, a deep learning-based infrared small target detection algorithm YOLO-NWD is proposed. According to the characteristics of small and medium targets in infrared images, multi-channel feature fusion image was used as the input of YOLO detection framework combined with image preprocessing method. Combined with SE module and ASPP module, feature weights are explored to improve feature utilization efficiency. Finally, the normalized Wasserstein distance (NWD) loss is used to replace the original IoU calculation loss to reduce the sensitivity of small target position deviation. The experimental results show that the algorithm proposed in this paper improves the accuracy by 2.5% and the recall rate by 4%.
- GONG Y,YUX, DING Y, Effective fusion factor in FPN for tiny object detection[C]/Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. USA: IEEE,2020: 1160-1168Google Scholar
- Konovalenko, I.; Maruschak, P.; Kozbur, H.; Brezinová, J.; Brezina, J.; Guzanová, A. Defectoscopic and Geometric Features of Defects That Occur in Sheet Metal and Their Description Based on Statistical Analysis. Metals 2021, 11, 1851.https://doi.org/10.3390/met11111851.Google Scholar
- Konovalenko, I.; Maruschak, P.; Brevus, V.; Prentkovskis, O. Recognition of Scratches and Abrasions on Metal Surfaces Using a Classifier Based on a Convolutional Neural Network. Metals 2021, 11, 549. https://doi.org/10.3390/met11040549.Google ScholarCross Ref
- Ren S, He K, Girshick R, Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2017, 39(6): 1137-1149.Google ScholarDigital Library
- J. Redmon, S. Divvala, R. Girshick and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779-788, doi: 10.1109/CVPR.2016.91.Google ScholarCross Ref
- Liu W , Anguelov D , Erhan D , SSD: Single Shot MultiBox Detector[J]. Springer, Cham, 2016.Google Scholar
- Zhou, X.; Ren, H.; Zhang, T.; Mou, X.; He, Y.; Chan, C.-Y. Prediction of Pedestrian Crossing Behavior Based on Surveillance Video. Sensors 2022, 22, 1467. https://doi.org/10.3390/s22041467Google Scholar
- He Y, Yang S, Zhou X, Lu XY. An Individual Driving Behavior Portrait Approach for Professional Driver of HDVs with Naturalistic Driving Data. Comput Intell Neurosci. 2022 Jan 22;2022:3970571. doi: 10.1155/2022/3970571. PMID: 35103055; PMCID: PMC8800608.Google Scholar
- Ju M, Luo J, Liu G, ISTDet: An efficient end-to-end neural network for infrared small target detection[J]. Infrared Physics & Technology, 2021, 114(7):103659.Google ScholarCross Ref
- Huang L, Dai S , Huang T , Infrared Small Target Segmentation with Multiscale Feature Representation[J]. Infrared Physics & Technology, 2021.Google Scholar
- Lichao, Zhang, Abel, Synthetic data generation for end-to-end thermal infrared tracking.[J]. IEEE transactions on image processing: a publication of the IEEE Signal Processing Society, 2018.Google Scholar
- WU Shuangchen,ZUO Zhengrong. Small target detection in infrared images using deep convolutional neural networks[J]. Journal of Infrared and Millimeter Waves,2019,38(3):371.Google Scholar
- Ding L, Xu X, Cao Y, Detection and tracking of infrared small target by jointly using SSD and pipeline filter[J]. Digital Signal Processing. 2021, 110: 102949.Google ScholarCross Ref
- Zhao Y, Infrared Dim and Small Target Detection Based on YOLOv3 in Complex Environment [J]. Areo Weaponry 2019, 26(06): 29-34.Google Scholar
- Zhang Ruzhen, Infrared target detection and recognition in complex scene[J]. Opto-Electronic Engineering. 2020, 47(10): 128-137.Google Scholar
- Cai, Y.; Li, D.; Zhou, X.; Mou, X. Robust Drivable Road Region Detection for Fixed-Route Autonomous Vehicles Using Map-Fusion Images. Sensors 2018, 18, 4158. https://doi.org/10.3390/s18124158Google Scholar
- Redmon J F A. YOLOv3: An Incremental Improvement[J]. ArXiv Preprint. 2018.Google Scholar
- K. H, X. Z, S. R, Deep Residual Learning for Image Recognition[C]. 2016.Google Scholar
- Szegedy C, Ioffe S, Vanhoucke V, Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning[J]. 2016.Google Scholar
- Yang X, Yan J , Ming Q , Rethinking Rotated Object Detection with Gaussian Wasserstein Distance Loss[J]. 2021.Google Scholar
- Hui Bingwei ,Song Zhiyong , A dataset for infrared detection and tracking of dim-small aircraft targets under ground / air background. 10.11922/csdata.2019.0074.zhGoogle Scholar
- Kisantal M , Wojna Z , Murawski J , Augmentation for small object detection[J]. 2019.Google ScholarCross Ref
Recommendations
Improved U-Net Network for Infrared Small Target Detection
AIPR '21: Proceedings of the 2021 4th International Conference on Artificial Intelligence and Pattern RecognitionInfrared small target detection is a key technology in infrared search and tracking systems. Despite the rapid development of computer vision in recent years, infrared small target detection based on deep learning is still few and far between due to the ...
Multiple Frames Based Infrared Small Target Detection Method Using CNN
ACAI '21: Proceedings of the 2021 4th International Conference on Algorithms, Computing and Artificial IntelligenceIn the field of infrared, the detection of dim small target has been a challenging topic. Especially in the complex background, a large number of false alarms might appear when using traditional detection methods, this cannot meet the dual requirements ...
Infrared dim and small target detection based on U-Transformer
AbstractInfrared dim and small target detection is a key technology for space-based infrared search and tracking systems. Traditional detection methods have a high false alarm rate and fail to handle complex background and high-noise ...
Highlights- Based on the U-shaped network, the feature extraction ability of the detector to the target is improved.
Comments