Detection of flying objects in images using the YOLOv4-CSP convolutional neural network model
The effectiveness of the YOLOv4-CSP convolutional neural network model in solving the problem of detecting objects moving in airspace is investigated. Images of flying objects of two classes were used as initial data for training and researching the convolutional neural network model: helicopter-type and aircraft-type unmanned aerial vehicles. Images of such objects were obtained in the optical and infrared wavelength ranges. Two datasets were formed from appropriately labeled source images with objects of these two classes. The first dataset was created from optical images, and the second from images obtained in the infrared wavelength range. The YOLOv4-CSP model was trained using training and validation samples from each dataset. Comprehensive studies of the effectiveness of the trained model were carried out using test samples from datasets. It is shown that the accuracy of detecting flying objects in optical images is higher than in images obtained in the infrared range, and the results for the speed of model calculation when analyzing optical and infrared images are close. Recommendations are given for the use of the YOLOv4-CSP model in computer vision systems for airspace monitoring. Contribution of the authors: the authors contributed equally to this article. The authors declare no conflicts of interests.
Keywords
computer vision system,
convolutional neural network YOLOv4-CSP,
detection of flying objects,
helicopter-type unmanned aerial vehicle,
aircraft-type unmanned aerial vehicleAuthors
| Nebaba Stepan G. | Tomsk Polytechnic University | stepanlfx@tpu.ru |
| Markov Nikolay G. | Tomsk Polytechnic University | markovng@tpu.ru |
Всего: 2
References
Zoev, I.V., Markov, N.G. & Ryzhova, S.E. (2019) Intelligent computer vision system for unmanned aerial vehicles for monitoring technological objects of oil and gas industry. Bulletin of the Tomsk Polytechnic University. Geo Assets Engineering. 330(11). pp. 34-49.
Alzubaidi, L., Zhang, J., Humaidi, A.J., Al-Dujaili, A., Duan, Y., Al-Shamma, O., Santamara, J., Fadhel, M.A., Al-Amidie, M. & Farhan, L. (2021) Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions. Journal of Big Data. 8(1). pp. 53.
Makarenko, S.I., Timoshenko, A.V. & Vasilchenko, A.S. (2020) Counter Unmanned Aerial Vehicles. Part 1. Unmanned aerial vehicle as an object of detection and destruction. Systems of Control, Communication and Security. 1. pp. 109-146.
Klekovkin, V.A. & Markov, N.G. (2023) Model of convolutional neural network LeNet5 for detection and classification of air space objects in images. Bulletin of the Tomsk Polytechnic University. Industrial Cybernetics. 1(2). pp. 11-16.
Nebaba, S.G. & Markov, N.G. (2022) Effectiveness of moving objects detecting and tracking in airspace by images in near infrared. Light & Engineering. 30(2). pp. 62-69.
Girshick, R., Donahue, J., Darrell, T. & Malik, J. (2014) Rich feature hierarchies for accurate object detection and semantic segmentation. Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 580-587.
Ren, S., He, K., Girshick, R. & Sun, J. (2015) Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 39(6). pp. 1137-1149.
Bochkovskiy, A., Wang, C.Y. & Liao, H.Y.M. (2020) YOLOv4: Optimal Speed and Accuracy of Object Detection. Journal arXiv, preprint arXiv: 2004.10934v1.
Redmon, J. & Ali, F. (2018) Yolov3: An incremental improvement. Journal arXiv, preprint arXiv:1804.02767.
Wang, C.Y., Mark Liao, H.Y., Yeh, I.H., Wu, Y.H., Chen, P.Y. & Hsieh, J.W. (2019) CSPNet: A New Backbone that can Enhance Learning Capability of CNN. Journal arXiv, preprint arXiv:. 1911.11929v1. 10.48550/arXiv. 1911.11929.
He, K., Zhang, X., Ren, S. & Sun, J. (2015) Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition. Journal arXiv, preprint arXiv. 1406.4729v4.
Wang, C.Y., Bochkovskiy, A. & Liao, H.Y.M. (2021) Scaled-YOLOv4: Scaling Cross Stage Partial Network. Proceedings of the IEEE/cvf conference on computer vision and pattern recognition.
