Application of CNN based image classification technique for oil spill detection
NOPR - NISCAIR Online Periodicals Repository
View Archive InfoField | Value | |
Title |
Application of CNN based image classification technique for oil spill detection
|
|
Creator |
Das, K
Janardhan, P Narayana, H |
|
Subject |
Convolutional Neural Network
Geographic Information System Image classification Oil spill Synthetic Aperture Radar |
|
Description |
5-14
Marine water pollution due to oil spills is a common threat to the environment worldwide because of its harmful impact on the economy and environment. Remote Sensing (RS) and Geographic Information Systems (GIS) are well-known tools for collecting satellite data which helps in remote oil spill identification. Synthetic Aperture Radar (SAR) images through various satellite missions are the mainly used data to identify oil spills. Many Artificial Neural Networks (ANN) and Machine Learning (ML) models integrated with RS and GIS have been originated and applied to identify and monitor oil spills. Deep Learning (DL) methods have recently become popular for their outstanding performance in research for image classification challenges, and the same is being used in the present study. An oil spill detection model using the Convolutional Neural Network (CNN) algorithm is presented in this work. CNN can extract features from a large dataset, and these features can be used to categorize images into different classes. The proposed model was compared with other existing models. The accuracy, precision, and recall achieved by this study are 99.06 %, 98.15 %, and 100 %, respectively. The proposed model outperformed the other existing work with an accuracy of 99.06 % and a precision of 98.15 %. |
|
Date |
2024-03-04T05:22:28Z
2024-03-04T05:22:28Z 2024-02 |
|
Type |
Article
|
|
Identifier |
2582-6727 (Online); 2582-6506 (Print)
http://nopr.niscpr.res.in/handle/123456789/63514 https://doi.org/10.56042/ijms.v52i01.5438 |
|
Language |
en
|
|
Publisher |
NIScPR-CSIR,India
|
|
Source |
IJMS Vol.52(01) [January 2023]
|
|