Home / Current Issue / Paper 1705630
Recognition of Jelly Fish by using CNN and SVM
Subject area: Science,Engineering and Technology · Area of research: Detection of fish
Abstract
The aim of this research is to devise a comprehensive approach for identifying jellyfish using Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs). To accomplish this, a deep learning CNN is utilized to extract image features by employing transfer learning with the MobileNetV2 architecture. The CNN is trained on a diverse dataset of various jellyfish species with the aid of an Image Data Generator for data augmentation. Additionally, a conventional SVM machine learning model is applied to evaluate resized and flattened image features. The SVM model is trained on a dataset comprising three different jellyfish types. The experimental outcomes demonstrate the effectiveness of both models, with the CNN achieving high accuracy on the training dataset, and the SVM exhibiting robust performance on a separate test dataset. Furthermore, a comparative analysis between the two models is conducted, highlighting their respective strengths and limitations. This combined approach offers a versatile solution for jellyfish recognition, combining the interpretability of SVMs with the representational power of CNNs.
Keywords
Jellyfish recognition, Convolutional Neural Networks, Support Vector Machines, Image classification, Comparative analysis.
How to cite this paper
@article{1705630,
author = {Jinal Babulal Gujar, Aman ShyamPrakash Mishra, Mithilesh Vishwakarma},
title = {Recognition of Jelly Fish by using CNN and SVM},
journal = {Iconic Research And Engineering Journals},
year = {2024},
volume = {7},
number = {9},
pages = {261-267},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1705630.pdf},
abstract = {The aim of this research is to devise a comprehensive approach for identifying jellyfish using Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs). To accomplish this, a deep learning CNN is utilized to extract image features by employing transfer learning with the MobileNetV2 architecture. The CNN is trained on a diverse dataset of various jellyfish species with the aid of an Image Data Generator for data augmentation. Additionally, a conventional SVM machine learning model is applied to evaluate resized and flattened image features. The SVM model is trained on a dataset comprising three different jellyfish types. The experimental outcomes demonstrate the effectiveness of both models, with the CNN achieving high accuracy on the training dataset, and the SVM exhibiting robust performance on a separate test dataset. Furthermore, a comparative analysis between the two models is conducted, highlighting their respective strengths and limitations. This combined approach offers a versatile solution for jellyfish recognition, combining the interpretability of SVMs with the representational power of CNNs.},
keywords = {Jellyfish recognition, Convolutional Neural Networks, Support Vector Machines, Image classification, Comparative analysis.},
month = {March},
}