Development of a Convolutional Neural Network for Classification of Type of Vessels

Ariel Victor do Nascimento, Marcus Pinto da Costa da Rocha, Valcir João da Cunha Farias, Miércio Cardoso de Alcântara Neto

Convolutional Neural Network, Artificial Intelligence; Deep Learning, Vessels

in this paper, was used a method which use concepts of intelligence artificial, machine learning, deep learning for Classification of Type of Vessels. With a technique from deep learning called Convolutional Neural Network (CNN) was applied to recognize images to identify the type of ship and to use the same method to identify if the vessel. The CNN projected with determined 5 layers, the first layer containing 32 neurons, the second layer with 64 neurons, the third layer with 128 neurons, the fourth layer with 512 neurons. Activation functions for these specified layers contain the ReLU function. The fifth and last layer is the output layer is the output layer, so the number of neurons is equal to the number of vessel type. In our study six classes were used, which are the vessel types, in this layer the activation function was the Softmax. The CNN generate satisfactory results, where could get results of prediction with all corrects answers to identify the ship.

Received: 16 Aug 2022, Receive in revised form: 18 Sep 2023, Accepted: 20 Oct 2022, Available online: 31 Jan 2023

10.22161/ijaers.101.21