Vol. 7, No 3: 40–44.

Computer Science and Informatics

 2022

Scientific article

UDK 004.89
pdf-version

Sabrina Sadiekh
bachelor's degree, Petrozavodsk State University
(Petrozavodsk, Russian Federation),
sad.sabrina.d@yandex.ru

Analysis of an approach to the construction of a fish population estimation system based on the Yolov4 neural network

Scientific adviser:
Alexandr R. Alexandrovich
Paper submitted on: 05/29/2022;
Accepted on: 06/12/2022;
Published online on: 10/01/2022.
 Abstract. The estimation of fish abundance is important for understanding the mechanisms of changes in marine ecosystems. For this purpose, video recording methods are increasingly used, the use of which, however, is limited by the complexity of manual image processing.
The article considers one of the possible approaches to the construction of an automatic population estimation system based on a neural network (NS). In the course of the work, a sufficiently high accuracy of counting the number of fish was obtained (correlation with manual counting of 0.996) and data on the patterns of tidal migrations of the three-spined stickleback in the White Sea coast were obtained.
Keywords: neural networks, White Sea, number, yolo, video image analysis

For citation: Sadiekh, S. Analysis of an approach to the construction of a fish population estimation system based on the Yolov4 neural network. StudArctic forum. 2022, 7 (3): 40–44.

References

Nadtochiy E. V., Savelyev P. D., Panitsina V. A., Demchuk A. S., Ivanov M. V., Ivanova T. S., Genelt-Yanovsky E. A., Laius D. L. Underwater interval surveys as a tool for studying the abundance and behavior of fish on the example of the water area of the Keret archipelago of the White Sea // Comprehensive research of the World Ocean. Materials of the VI All-Russian scientific conference of young scientists. Moscow, 2021. P. 288—289.

Tashbaev D. U., Sadiekh S. A., Nadtochiy E. V., Genelt-Yanovsky E. A., Demchuk A. S., Ivanova T. S., Ivanov M. V., Laius D. L. Use video recording for fish studies: automatic counting of three-spined stickleback on video images in the coastal zone of the White Sea // All-Russian scientific and practical conference with international participation "Arctic ecosystems: conservation and sustainable development". Murmansk: MAGU, 2021 (article in press).

Chien-Yao Wang, Alexey Bochkovskiy, Hong-Yuan Mark Liao. Scaled-YOLOv4: Scaling Cross Stage Partial Network, Nov 16, 2020

Chien-Yao Wang, Alexey Bochkovskiy, Hong-Yuan Mark Liao. YOLOv4: Optimal Speed and Accuracy of Object Detection, Git code, 2020, URL: https://github com.teameo.ca/AlexeyAB/darknet (accessed 06/08/2022)

Konovalov D. A., Saleh A., Bradley M., Sankupellay M., Marini S., Sheaves M. Underwater detection with weak multi-domain supervision. International Joint Conference on Neural Networks (IJCNN), IEEE. - 2019. - P. 1-8;

Tzutalin, LabelImg, Git code, 2015, URL: https://github.com/tzutalin/labelImg (accessed 06/08/2022).

Displays: 397; Downloads: 81;