Vol. 7, No 4: 35–40.

Computer Science and Informatics

 2022

Scientific article

UDK 004.89
pdf-version

Maksim P. Pavlov
bachelor’s degree, Petrozavodsk State University
(Petrozavodsk, Russian Federation),
maksim_pavlov_2003@list.ru

Comparison of YOLOv5’s models for the rainbow trout detection in a video stream

Scientific adviser:
Alexey G. Marakhtanov
Paper submitted on: 11/11/2022;
Accepted on: 11/20/2022;
Published online on: 12/25/2022.
 Abstract. The diversity of models and training parameters makes it difficult to find an optimal solution for the object detection problem. In this article the models of YOLOv5 are considered. For each of them, the key learning metrics are measured and the Pareto front of optimal solutions is constructed using the rectangle method. Based on this, the best model in terms of accuracy and speed of rainbow trout detection is selected. The author concludes that the training of models depends not only on their architecture, the hardware capabilities of the system, but also on the number of detection classes.
Keywords: neural networks, YOLOv5, rainbow trout detection, optimal set, Pareto front

For citation: Pavlov, M. P. Comparison of YOLOv5’s models for the rainbow trout detection in a video stream. StudArctic forum. 2022, 7 (4): 35–40.

References

COCO Annotator. URL: https://github.com/jsbroks/coco-annotator?ref=madewithvuejs.com (accessed: 29.09.2022).

Deng J. et al. Imagenet: A large-scale hierarchical image database. 2009 IEEE conference on computer vision and pattern recognition, 2009, P. 248—255.

Fish Dataset. URL: https://public.roboflow.com/object-detection/fish (accessed: 29 September 2022).

Kuznetsova A. et al. The open images dataset v4. International Journal of Computer Vision, 2020, V.128, I.7, p. 1956-1981.

Letov A. M. Dinamika poleta i upravlenie [Flight dynamics and control]. – Moscow, Nauka Publ., 1969. (in Russian)

Muxanov K. A. Algoritmy` mnogokriterial`noj optimizacii v zadachax vy`bora parametrov sistemy` upravleniya mexatronny`mi modulyami [Multicriterian optimisation algorithms in mechatronic module control system parameter selection problems]. Scientific journal "Student Forum", 2017, no.10, pp.12-15. (in Russian)

Pedersen M. et al. Detection of marine animals in a new underwater dataset with varying visibility. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2019, p. 18-26.

Ulucan O., Karakaya D., Turkan M. A Large-Scale Dataset for Fish Segmentation and Classification. 2020 Innovations in Intelligent Systems and Applications Conference (ASYU), 2020, P. 1—5.

YOLOv5 Documentation. Available at: https://docs.ultralytics.com/ (accessed: 29 September 2022).

Yu Yongcan, Zhao Jianhu, Gong Quanhua, Huang Chao, Zheng Gen, Ma Jinye Real-Time Underwater Maritime Object Detection in Side-Scan Sonar Images Based on Transformer-YOLOv5. Remote Sensing, 2021, V. 18, I. 13, P. 1—28.

Displays: 535; Downloads: 84;