DEEP LEARNING ARCHITECTURES FOR VISUAL RECOGNITION OF SELECTED TYPES OF EXPLOSIVE ORDNANCE: COMPARATIVE ANALYSIS AND SYSTEM IMPLEMENTATION

Volodymyr Hrabovskyi, Oleksii Hryhorashyk

Abstract


Background. In the context of significant contamination of Ukraine's territory with explosive ordnance (EO) resulting from the Russian Federation's aggression and over 11 years of ongoing hostilities, coupled with limited resources for detection and demining, the implementation of innovative technologies for identifying and recognizing such objects is critical. Therefore, the application of artificial intelligence-based methods for the detection and recognition of explosive devices is both relevant and essential.

Materials and Methods. This article explores the potential of Deep Neural Networks (DNNs) and computer vision for automating EO detection. The study reviews the theoretical foundations and practical applications of Convolutional Neural Networks (CNNs) in object recognition tasks. The primary focus is on utilizing modern deep learning methods to identify explosive devices, specifically technologies employing single-stage (SSD – Single Shot MultiBox Detector, and YOLO – You Only Look Once) and two-stage (R-CNN, Regions-based  Convolutional Neural Networks) approaches for object detection and identification. Key aspects of using single- and two-stage DNN architectures for visual object recognition are investigated, comparing the strengths and weaknesses of both approaches and examining ways to improve them to enhance detection efficiency.

Results and Discussion. Based on the conducted analysis, models were developed using single-stage (SSD and YOLOv8) and two-stage (Faster R-CNN) object identification technologies. These models were trained on an original dataset created from internet-sourced images, which included eight common types of EO found in combat zones and de-occupied territories. Testing results of the trained models demonstrated the superiority of the YOLOv8 architecture for explosive device detection and recognition, which was subsequently used to develop a web server for explosive ordnance recognition.

Conclusions. A project was developed to train EO recognition model architectures based on two-stage and single-stage approaches. Using these models, a web server with a REST API (Representational State Transfer Application Programming Interface) was created for EO recognition in both static images and real-time video streams. The proposed system can facilitate demining operations and reduce risks to civilian populations.

Keywords: deep learning, visual recognition, explosive ordnance, SSD (Single Shot MultiBox Detector), YOLOv8, Faster R-CNN


Full Text:

PDF

References


[1] Interactive map of areas potentially contaminated with explosives. (ukr) URL: https://mine.dsns.gov.ua/

[2] What percentage of Ukraine’s territories are mined? (ukr). URL: https://tsn.ua/ukrayina/skilki-vidsotkiv-teritoriy-ukrayini-zaminovani-klimenko-prigolomshiv-cifroyu-2550241.html

[3] Over 144,000 sq km of Ukraine are considered potentially mined – new data from the Ministry of Internal Affairs. (ukr). URL: https://suspilne.media/781039-ponad-144-tisaci-kv-km-ukraini-vvazautsa-potencijno-zaminovanimi-novi-dani-mvs/

[4] Demining Ua – Rozminuvannya Ukraiyy (ukr). URL: https://deminingua.com/rozminuvannya-ukrayini

[5] Szeliski, R. (2022). Computer vision: Algorithms and applications (2nd ed.). Springer. https://doi.org/10.1007/978-3-030-34372-9

[6] Zou, Z., Chen, K., Shi, Z., Guo, Y., & Ye, J. (2023). Object detection in 20 years: A survey. Proceedings of the IEEE, 111(3), 257–276. https://doi.org/10.48550/arxiv.1905.05055

[7] Lamichhane, B. R., Srijuntongsiri, G., & Horanont, T. (2025). CNN based 2D object detection techniques: A review. Frontiers in Computer Science, 7, 1437664. https://doi.org/10.3389/fcomp.2025.1437664

[8] Edozie, E., et al. (2025). Comprehensive review of recent developments in visual object detection based on deep learning. Artificial Intelligence Review, 58, 277. https://doi.org/10.1007/s10462-025-11284-w

[9] Voulodimos, A., Doulamis, N., Doulamis, A., & Protopapadakis, E. (2018). Deep learning for computer vision: A brief review. Computational Intelligence and Neuroscience, 2018, Article 7068349. https://doi.org/10.1155/2018/7068349

[10] Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). ImageNet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84–90. https://doi.org/10.1145/3065386

[11] Sermanet, P., Eigen, D., Zhang, X., Mathieu, M., Fergus, R., & LeCun, Y. (2014). OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks. In International Conference on Learning Representations (ICLR 2014). https://arxiv.org/abs/1312.6229

[12] Szegedy, C., Toshev, A., & Erhan, D. (2013). Deep neural networks for object detection. In C. J. Burges, L. Bottou, M. Welling, Z. Ghahramani, & K. Q. Weinberger (Eds.), Advances in neural information processing systems (Vol. 26). Curran Associates, Inc. https://proceedings.neurips.cc/paper/2013/file/f7cade80b7cc92b991cf4d2806d6bd78-Paper.pdf.

[13] Uijlings, J. R. R., van de Sande, K. E. A., Gevers, T., & Smeulders, A. W. M. (2013). Selective search for object recognition. International Journal of Computer Vision, 104(2), 154–171. https://doi.org/10.1007/s11263-013-0620-5

[14] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587). https://doi.org/10.1109/CVPR.2014.81

[15] He, K., Zhang, X., Ren, S., & Sun, J. (2015). Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE transactions on pattern analysis and machine intelligence, 37(9), 1904-1916. https://doi.org/10.1109/TPAMI.2015.2389824

[16] Girshick, R. (2015). Fast R‑CNN. In Proceedings of the IEEE international conference on computer vision (pp. 1440-1448). https://doi.org/10.1109/ICCV.2015.169

[17] Ren, S., He, K., Girshick, R., & Sun, J. (2016). Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence, 39(6), 1137-1149. https://doi.org/10.1109/TPAMI.2016.2577031

[18] Lin, T. Y., Dollár, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017). Feature pyramid networks for object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 2117–2125). https://doi.org/10.1109/CVPR.2017.106

[19] He, K.M., Gkioxari, G., Dollár, P. and Girshick, R. (2017) Mask R-CNN. IEEE International Conference on Computer Vision (ICCV), Venice, 22-29 October 2017, 386-397. https://doi.org/10.1109/ICCV.2017.322

[20] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real‑time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 779–788). IEEE. https://doi.org/10.1109/CVPR.2016.91

[21] Redmon, J., & Farhadi, A. (2017). YOLO9000: Better, faster, stronger. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7263–7271). IEEE. https://doi.org/10.1109/CVPR.2017.690

[22] Redmon, J., & Farhadi, A. (2018). YOLOv3: An incremental improvement. arXiv preprint arXiv:1804.02767. https://doi.org/10.48550/arXiv.1804.02767

[23] Bochkovskiy, A., Wang, C.-Y., & Liao, H.-Y. M. (2020). YOLOv4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934. https://arxiv.org/abs/2004.10934

[24] Jocher, G. (2020). YOLOv5 by Ultralytics (Version 7.0) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.3908559

[25] Wang, C. Y., Bochkovskiy, A., & Liao, H. Y. M. (2023). YOLOv7: Trainable bag-of-freebies for real-time object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7701–7711).
https://doi.org/10.48550/arXiv.2207.02696

[26] Ultralytics. (2023). YOLOv8 Official Documentation. Retrieved from https://docs.ultralytics.com/yolov8/

[27] Wang, C. Y., Liao, H. Y. M., & Yeh, I. H. (2024). YOLOv9: Learning what you want to learn using programmable gradient information. arXiv preprint arXiv:2402.13616. https://arxiv.org/abs/2402.13616

[28] Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.‑Y., & Berg, A. C. (2016). SSD: Single shot multibox detector. In B. Leibe, J. Matas, N. Sebe, & M. Welling (Eds.), Computer vision – ECCV 2016. Lecture Notes in Computer Science (Vol. 9905, pp. 21–37). Springer, Cham.
https://doi.org/10.1007/978-3-319-46448-0_2

[29] Lin, T. Y., Goyal, P., Girshick, R., He, K., & Dollár, P. (2017). Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision (ICCV) (pp. 2977–2985). IEEE. https://doi.org/10.48550/arXiv.1708.02002

[30] Tan, M., Pang, R., & Le, Q. V. (2020). EfficientDet: Scalable and efficient object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 10781–10790). IEEE. https://doi.org/10.48550/arXiv.1911.09070

[31] Tian, Z., Shen, C., Chen, H., & He, T. (2019). FCOS: Fully convolutional one‑stage object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) (pp. 9627–9636). IEEE. https://doi.org/10.48550/arXiv.1904.01355

[32] Domingos, P. (2015). The master algorithm: How the quest for the ultimate learning machine will remake our world. Basic Books.

[33] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT Press.

[34] Bishop, C. M. (2006). Pattern recognition and machine learning. Springer.

[35] CVAT.io – Open Data Annotation Platform. URL: https://www.cvat.ai/

[36] mAP (mean Average Precision) for Object Detection, URL: https://jonathan-hui.medium.com/map-mean-average-precision-for-object-detection-45c121a31173

[37] Shobaki, W. A., & Milanova, M. (2025). A comparative study of YOLO, SSD, Faster R-CNN, and more for optimized eye-gaze writing. Sci, 7(2), 47. https://doi.org/10.3390/sci7020047




DOI: http://dx.doi.org/10.30970/eli.34.7

Refbacks

  • There are currently no refbacks.