V.A. Pavlov – Leading Engineer, 

Higher School of Applied Physics and Space Technologies of Peter The Great St. Petersburg Polytechnic University E-mail: pavlov_va@spbstu.ru

This article discusses the problems of detecting and recognizing artificial space and ground objects on images. The approaches based on convolutional neural networks of YOLO family for the detection and recognition of images of small-sized artificial space objects against the background of starry sky and ground objects on aerial photographs have been considered. Datasets consisting of images of artificial space objects against the background of starry sky and ground objects on aerial photographs have been developed. The results of the operation of convolutional neural networks on generated datasets have been obtained. The studied versions of trained neural networks showed high reliability of detection and recognition of objects against complex background. The researches were shown that an increase in the number of classes, complexity of scene model and required reliability of recognition leads to some complication of the architecture of the convolutional neural network.

1. Bakir N., Pavlov V., Zavjalov S., Volvenko S., Gumenyuk A., Rethmeier M. Development of a novel optical measurement technique to investigate the hot cracking susceptibility during laser beam welding. Welding in the World. 2019. T. 63. № 2. S. 435−441.
2. Bakir N., Pavlov V., Zavjalov S., Volvenko S., Gumenyuk A., Rethmeier M. Novel metrology to determine the critical strain conditions required for solidification cracking during laser welding of thin sheets. Journal of Physics: Conference Series. 2018. № 1109. S. 1−9.
3. Bobrovsky A., Galeeva M., Morozov A., Pavlov V., Tsytsulin A. Automatic detection of objects on star sky images by using the convolutional neural network. Journal of Physics: Conference Series. 2019. № 1236. S. 1−6.
4. Obukhova N.A., Timofeev B.C. Zakhvat i soprovozhdenie ob’ektov v avtomatizirovannoi sisteme. Materialy IV Mezhdunar. nauch.praktich. konf. «Elektronnye sredstva i sistemy upravleniya». 12−14 oktyabr 2005. Tomsk. S. 277−280. (in Russian)
5. Obukhova N.A. Vektory opticheskogo potoka v zadachakh segmentatsii i soprovozhdeniya podvizhnykh ob’ektov. Izvestiya VUZov Rossii. Ser. Radioelektronika. 2006. № 2. S. 42−51. (in Russian)
6. Lukyanitsa A.A., Shishkin A.G. Tsifrovaya obrabotka videoizobrazhenii. M.: Ai–Es–Es Press. 2009. 518 s. (in Russian)
7. Khaikin S. Neironnye seti. Polnyi kurs. M.: Izdatelskii dom Vilyams. 2006. 1104 s. (in Russian)
8. Dalal N., Triggs B. Histograms of oriented gradients for human detection. IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), San Diego, CA, USA. 2005. T. 1. S. 886−893.
9. Viola P., Jones M.J. Rapid object detection using a boosted cascade of simple features. Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2001), Kauai, HI, USA. 2001. S. 511−518.
10. Felzenszwalb P.F., Girshick R.B., McAllester D. et al. Object detection with discriminatively trained part based models. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2010. T. 32. № 9. S. 1627−1645.
11. Girshick R., Donahue J., Darrell T., Malik J. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation. IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH. 2014. S. 580−587.
12. Girshick R. Fast R-CNN. IEEE International Conference on Computer Vision (ICCV). Santiago. 2015. S. 1440−1448.
13. Ren S., He K., Girshick R., Sun J. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2017. T. 39. № 6. S. 1137−1149.
14. Redmon J., Divvala S., Girshick R., Farhadi A. You Only Look Once: Unified, Real-Time Object Detection. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV. 2016. S. 779−788.
15. Redmon J., Farhadi A. YOLO9000: better, faster, stronger. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI. 2017. S. 6517−6525.
16. Redmon J., Farhadi A. Yolov3: An incremental improvement. arXiv, 2018. Preprint, arXiv:1804.02767v1.
17. Bobrovskii A.I., Morozov A.V., Tsytsulin A.K. i dr. Obnaruzhenie ob’ektov na zvezdnom fone. Voprosy radioelektroniki. Ser. Tekhnika televideniya. 2016. № 2. S. 29−38. (in Russian)
18. Tsytsulin A.K., Morozov A.V., Bobrovskii A.I. Baskakova Yu.V. i dr. Klassifikatsiya malorazmernykh izobrazhenii kosmicheskikh ob’ektov po priznakam dvizheniya s pomoshchyu obuchaemogo algoritma. Voprosy radioelektroniki. Seriya Tekhnika televideniya. 2018. № 3. S. 72−80. (in Russian)
19. Lin T., Goyal P., Girshick R. et al. Focal loss for dense object detection. arXiv, 2017. Preprint arXiv:1708.02002.
20. Lin T., Dollár P., Girshick R., He K,. Hariharan B., Belongie S. Feature Pyramid Networks for Object Detection. arXiv, 2017. preprint arXiv:1612.03144.
21. [Elektronnyi resurs]. Rezhim dostupa: captain-whu.github.io/DOTA/index.html (data obrashcheniya: 1.09.2019).
22. [Elektronnyi resurs].Rezhim dostupa:vision.cse.psu.edu/data/vividEval/datasets/datasets.html (data obrashcheniya: 1.09.2019).
23. [Elektronnyi resurs].Rezhim dostupa: kaggle.com (data obrashcheniya: 1.09.2019).
24. [Elektronnyi resurs].Rezhim dostupa: google.ru/maps (data obrashcheniya: 1.09.2019).
25. Everingham M., Van Gool L., Williams C.K.I. et al. The Pascal Visual Object Classes (VOC) Challenge. International Journal of Computer Vision. 2010. № 2. S. 303−338.
26. Lin M., Chen Q., Yan S. Network In Network. arXiv, 2013. Preprint arXiv: arXiv:1312.4400v3.
27. Kristan M., Leonardis A., Matas J. et al. The sixth visual object tracking VOT2018 challenge results. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2019. T. 11129. S. 3−53.