THE ROLE OF COMPUTER VISION IN THE MODERN WORLD: ACHIEVEMENTS, CHALLENGES AND PROSPECTS

Authors

DOI:

https://doi.org/10.35546/kntu2078-4481.2024.4.40

Keywords:

object identification, computer vision, convolutional neural networks, pattern recognition

Abstract

This research paper proposes a new approach to object identification in static visual scenes that combines advanced machine learning and image processing techniques. Due to the two-stage structure, the proposed identification method provides high accuracy and speed of object detection in various lighting, distortion and noise conditions. The advantages of the new approach include high performance, efficient use of resources and scalability for different application areas. The purpose of the study is to increase the accuracy and speed of object identification in static visual scenes, resistance to various types of noise and distortions. The research involves the optimization of image processing and machine learning algorithms that allow for the efficient selection and classification of objects in conditions of limited resources and variability of visual scenes. The methodology consists in a comparative analysis of existing object identification methods, including both twostage and one-stage approaches. Their advantages and limitations are revealed. An empirical study and evaluation of the effectiveness and accuracy of various approaches to the detection and identification of objects was carried out. The scientific novelty of the results obtained in the work is the combination of one-stage and two-stage detection of objects using RefineDet, which allows to improve the accuracy and speed of object detection, compared to traditional one-stage methods. Conclusions. RefineDet is a powerful and efficient method for detecting objects in images. Its uniqueness lies in the combination of the advantages of one-stage and two-stage approaches, which ensures high accuracy and speed of work. This makes it suitable for use in various real-world applications where fast and accurate object identification is required.

References

W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, and C. Y. Fu. SSD: Single shot multibox detector, in Proc. Eur. Conf. Comput. Vis., Amsterdam, The Netherlands, Oct. 2016, pp. 11–14.

R. Girshick, J. Donahue, T. Darrell, and J. Malik. Rich feature hierarchies for accurate object detection and semantic segmentation, in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., Jun. 2014, pp. 23–28

Sana Pavan Kumar Reddy. Cusp Pixel Labelling Model for Objects Outline Using R-CNN. 2021. URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9667355 (дата звернення: 01.11.2024).

S. Ren, K. He, R. Girshick, J. Sun. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Trans. Pattern Anal. Mach. Intellegence, 39 (2017), pp. 1137-1149.

K. He G. Gkioxari P. Dollár R. Girshick. Mask R-CNN Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV) Venice, Italy 22–29 October 2017.

J. An, A. Mikhaylov, K. Kim. Machine learning approach in heterogeneous group of algorithms for transport safetycritical system. Appl. Sci., 10. 2020.

S.M. Abbas S.N. Singh Region-based object detection and classification using faster R-CNN Proceedings of the 2018 4th International Conference on Computational Intelligence & Communication Technology (CICT) Ghaziabad, India 9–10 February 2018.

Jiaxu Leng, Ying Liu, Xinbo Gao, Senior Member. Selective region enlargement network for fast object detection in high resolution images. URL: https://doi.org/10.1016/j.neucom.2021.08.015 (дата звернення: 01.11.2024).

S. Ren, K. He, R. Girshick and J. Sun, "Faster R-CNN: Towards real-time object detection with region proposal networks", IEEE Trans. Pattern Anal. Mach. Intell., vol. 39, no. 6, pp. 1137-1149, Jun. 2017.

J. Redmon, S. Divvala, R. Girshick and A. Farhadi, "You only look once: Unified real-time object detection", Proc. IEEE Conf. Comput. Vis. Pattern Recognit., pp. 779-788, Jun. 2016.

W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y. Fu, et al., "SSD: Single shot multibox detector" in Computer Vision–ECCV, Cham, Switzerland:Springer, pp. 21-37, 2016.

Kai Shuang, Zhiheng Lya, Jonathan Loo, Wentao Zhang. Scale-balanced loss for object detection. URL: https://doi.org/10.1016/j.patcog.2021.107997 (дата звернення: 04.11.2024).

Published

2024-12-30