Comparison and Analysis of CNN based Underwater Aquatic Products Recognition Methods
DOI:
https://doi.org/10.53469/jrse.2024.06(09).04Keywords:
Underwater images, Underwater aquatic products recognition, CNNAbstract
Underwater aquatic products are naturally cultured in seawater exceeding two meters to obtain excellent quality. The catching of these aquatic products mostly relies on professional fishermen, which consumes a lot of manpower and material resources. In recent years, underwater fishing robots have emerged, but due to inaccurate positioning of Underwater aquatic products, the fishing efficiency is not satisfactory. Based on CNN, underwater aquatic product recognition methods were researched. Firstly, an underwater aquatic products recognition dataset containing 5443 aquatic product images was constructed based on the training data provided by the National Underwater Robot Competition - Underwater Object Detection Competition. Subsequently, SSD, Faster RCNN, YOLO V5, and YOLO V8 were used to recognize underwater aquatic products on the above dataset and the recognition performance of various methods was compared. The experimental results show that YOLO V8 has the most ideal recognition performance, with an mAP value of 0.862.
References
CHEN C H, LEE J D, LIN M C. Classification of underwater signals using wavelet transforms and neural networks[J]. Mathematical and Computer Modelling, 1998, 27(2):47-60.
GUO J, MA J F, WANG A. Research on classification of side-scan sonar images based on SVM algorithm and GLCM[J]. Surveying and Mapping and Spatial Geographic Information, 2015, 38(3):60-63.
ZHU B Y, WANG X, CHU Z, et al. Active learning for recognition of shipwreck target in side-scan sonar image[J]. Remote Sensing, 2019, 11(3):1030243.
DONG J N, YANG M, XIE Z R, et al. Overview of underwater image target detection data set and detection algorithm [J]. Journal of marine technology, 2022, 41(5): 60-72.
Ren S, He K, Girshick R, et al. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 39(6):1137-1149.
Joseph Redmon, Ali Farhadi. YOLOv3: An Incremental Improvement [EB/OL]. [2018], https://alumni.soe. ucsc.edu/~czczycz/src/YOLOv3.pdf.
HUANG C X, LIU P T. Research on underwater target recognition method based on CA-YOLOv5s [J]. Journal of marine technology, 2024, 43(03):17-24.
Joseph Redmon. YOLO V5[EB/OL]. [2020-05-18]. https://github.com/ultralytics/yolov5
QIANG W, HE R Y, GUO Y J, et al. Research on Underwater Target Detection Algorithm Based on Improved SSD [J]. Journal of northwestern polytechnical university, 2020, 38(4): 747-754.
Liu W, Anguelov D, Erhan D, et al. SSD: Single Shot MultiBox Detector[J]. 14th European Conference on Computer Vision (ECCV), 2015, (9905):21-37.
Simonyan K, Zisserman A. 2014. Very Deep Convolutional Networks for Large-Scale Image Recognition [J]. Computer Science.
He K, Zhang X, Ren S, et al. 2016. Deep Residual Learning for Image Recognition [C]. IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
LI H J, LIAN C, GUAN L C, et al. Study on identification and distribution of organisms in deep-sea cold spring area based on R-FCN_ResNet [J]. Marine science, 2020, 44(9): 54-62.
Ultralytics. YOLO V8 [EB/OL]. [2023-02-03]. https://github.com/ultralytics/ultralytics.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2024 Liqiong Lu, Dong Wu, Liuyin Wang
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
