Navigation safety is an important issue to maintain the well-being and integrity
of passengers and cargo. There are many rules to follow to assess safety, certifiers
and classifiers are responsible for ensuring compliance with all these rules that ensure
the integrity of vessels, however, this is not enough. The Administrative Inquiry of
Naval Accidents and Facts (IAFN), a document prepared by the Brazilian Navy, collects
information and creates a database to show how many accidents occur in Brazil by region,
which are defined as Naval Districts (DN). The 4th Naval District, in which the state of
Par´a is located, was the first in accidents that occurred in 2020 and the third in 2021. Due
to these accidents, concepts of artificial intelligence, machine learning and deep learning
were used. applied in this area. In order to assist in this process, this work proposes to
develop an application using Convolutional Neural Network (CNN) for image recognition
(Vessels and Disc of plimsoll). In this sense, a Convolutional Neural Network (CNN)
learning technique was used, which allows identifying the type of vessel through a bank
of images provided, the same method was applied to identify if there is a risk of accident
with the vessel through analysis of disk images of plimsoll. To carry out the training of
the CNNs, six different network architectures were evaluated with: changing the number
of filters in each convolutional layer; variation in the amount of convolutional layers and;
use of transfer of learning from the VGG-16 network with the fine-tuning technique. The
results achieved in this work are promising and demonstrate the feasibility of using the
Convolutional Neural Network as a method for identifying the images of vessels as the
disk of plimsoll).
|
