Research of YOLO Architecture Models in Book Detection

Maria Kalinina; Pavel Nikolaev

Информационные технологии интеллектуальной поддержки принятия решений, Информационные технологии интеллектуальной поддержки принятия решений 2020

Maria Kalinina, Pavel Nikolaev

Изменена: 2025-02-20

Аннотация

—Deep neural networks are widely used in different fields of human activity, including spheres which are connected with large amount of operations such as data obtaining and processing information from the outside world. This article deals with the creation of the deep convolutional neural network based on the YOLO architecture for book detection in real time. The architecture chosen as the basis of the neural network possesses a number of advantages which make it highly competitive with other models, so it can be considered as the most suitable option for the creation of deep neural network for object detection. Creation of the original dataset and the deep neural network training are described. Several variants of neural networks based on the YOLO architecture are discussed and the results of their comparison are shown. The results obtained during the training of a deep neural network allow us to use it as a basis for further development of the application.

Ключевые слова

image recognition; object detection; computer vision; machine learning; artificial neural networks; deep learning; convolutional neural networks

Литература

[1] J. Redmon, S. Divvala, R. Girshick, A. Farhadi. “You only look once: Unified, real-time object detection,” Proc. 2016 IEEE Conference on Computer Vision and Pattern Recognition, pp. 779–788, 2016.

[2] J. Redmon, A. Farhady. “Yolo9000: Better, faster, stronger,” Proc. 2017 IEEE Conference on Computer Vision and Pattern Recognition, pp. 6517–6525, 2017.

[3] J. Redmon, A. Farhady. “YOLOv3: An Incremental Improvement,” Proc. 2018 IEEE Conference on Computer Vision and Pattern Recognition, 2018.

[4] Y. Chang , A. Anagaw, L. Chang, Y. Chun Wang, C. Hsiao, W. Lee. “Ship Detection Based on YOLOv2 for SAR Imagery,” Remote Sensing – Open Access Journal, 2 April 2019.

[5] C. Rajesh Babu, G. Anirudh. “Vehicle Traffic Analysis Using Yolo,” Eurasian Journal of Analytical Chemistry, vol. 13, pp. 345–350, 2018.

[6] A. V. Devyatkin, D. M. Filatov. “Neural network traffic signs detection system development,” Proc. International conference on soft computing and measurement, vol. 1, pp. 189-192, 2019 [Nejrosetevaya sistema obnaruzheniya znakov dorozhnogo dvizheniya].

[7] C. Zhang, C.C. Chang, M. Jamshidi. “Bridge Damage Detection using a Single-Stage Detector and Field Inspection Images,” Computer Society, 8 April 2018.

[8] K.S. Kurochka, T.V. Luchsheva, K.A. Panarin. “Localization of human percentages on X-ray images with use of Darknet YOLO,” Doklady BGUIR, Vol. 113, Nо. 3, pp. 32–38, 2018 [Lokalizaciya pozvonkov cheloveka na rentgenovskih izobrazheniyah s ispol'zovaniem DARKNET YOLO].

[9] “YOLO: Real-Time Object Detection”. Source: https://pjreddie.com/darknet/yolo/

[10] T.C. Arlen. “Understanding the mAP Evaluation Metric for Object Detection,” Medium, 1 March 2018.