Sea Fog Identification from GOCI Images Using CNN Transfer Learning Models

Ho-Kun Jeon,Chan-Su Yang,Seungryong Kim,Jonathan Edwin

doi:10.3390/electronics9020311

Abstract

This study proposes an approaching method of identifying sea fog by using Geostationary Ocean Color Imager (GOCI) data through applying a Convolution Neural Network Transfer Learning (CNN-TL) model. In this study, VGG19 and ResNet50, pre-trained CNN models, are used for their high identification performance. The training and testing datasets were extracted from GOCI images for the area of coastal regions of the Korean Peninsula for six days in March 2015. With varying band combinations and changing whether Transfer Learning (TL) is applied, identification experiments were executed. TL enhanced the performance of the two models. Training data of CNN-TL showed up to 96.3% accuracy in matching, both with VGG19 and ResNet50, identically. Thus, it is revealed that CNN-TL is effective for the detection of sea fog from GOCI imagery.

Highlights

Sea fog is one of the major reasons for maritime accidents in Korea because of poor visibility [1,2,3]
This paper introduces the sea fog identification method using Geostationary Ocean Color Imager (GOCI) with a Convolutional Neural Network Transfer Learning (CNN-TL) model that has been trained with an ImageNet dataset [25]
In this study we conducted an identification experiment, varying whether TL is applied, using only 100 images for training per class, in five kinds of band combination, setting 20 epochs. When it comes to TL, VGG19 and ResNet50 were remarkably enhanced for sea fog identification

Summary

Introduction

Sea fog is one of the major reasons for maritime accidents in Korea because of poor visibility (less than 1 km) [1,2,3]. It deters navigators from keeping a lookout for surrounding ships and obstacles [4]. The number of coastal meteorological stations is not enough to represent sea fog across the oceans [6]. In this regard satellite remote sensing technology could be a decent method for monitoring sea fog with its wide coverage

Methods

Results

Discussion

Conclusion