Application of SAR images in Iceberg Classification by using ConvNet-2

Iceberg areas are not safe for transportation because based on climate changes icebergs are melting and not showing a dangerous way in sea areas. Based on the vision we can’t identify all the icebergs in the ocean area. So, by using a Deep learning algorithm we can classify icebergs through satellite images. In the past decades, several machine learning algorithms are implemented for classification of the images. But our aim is to implement an application to classify the iceberg by using synthetic-aperture radar (SAR) images which are available at the Kaggle repository. The Data set was from the Statoil C-CORE East Coast of Canada. Here we classify the icebergs by using remotely sensed data. For this data, the Convolutional Neural Network is used for image classification and extraction of the features of images deeply. The CNN algorithm was implemented on the SAR images and achieved 99.8% training and 89.5% of validation accuracy with high time consumption when training the dataset.