Eventos Anais de eventos
COBEM 2017
24th ABCM International Congress of Mechanical Engineering
Deep Learning Approach Based on Convolutional Neural Networks for Image Processing Applications
Submission Author:
Leandro dos Santos Coelho , PR , Brazil
Co-Authors:
João Sauer, Marco Boaretto, Edson Gnatkovski Gruska, arthur canciglieri, Gabriel Herman Bernardim Andrade, Leandro dos Santos Coelho
Presenter: Gabriel Herman Bernardim Andrade
doi://10.26678/ABCM.COBEM2017.COB17-1471
Abstract
Machine learning has been successfully applied to a wide variety of knowledge fields ranging from information retrieval, data mining, and speech recognition, to computer graphics, visualization, and human-computer interaction. The general focus of machine learning is the representation of the input data and generalization of the learnt patterns for use on future unseen data. The goodness of the data representation has a large impact on the performance of machine learners on the data: a poor data representation is likely to reduce the performance of even an advanced, complex machine learner, while a good data representation can lead to high performance for a relatively simpler machine learner. In this context, deep learning has received significant attention recently as a promising solution to many complex problems in the Artificial Intelligence and signal processing fields related to processing of unstructured data. Among different types of deep artificial neural networks, convolutional neural networks (CNNs) have been most extensively studied. CNNs demonstrate superior performance when compared to other machine learning methods in the object detection and recognition applications. Each stage in a CNN is composed of a filter bank, some non-linearities, and feature pooling layers. With multiple stages, a CNN can learn multi-level hierarchies of features. The contribution of this paper is related to an overview of CNN applications and its key benefits related to image processing and computer vision. Furthermore, an image processing case study related to Mechatronics and Automation in the automotive field is evaluated using a CNN design approach.
Keywords
machine learning, Deep learning, Convolutional neural networks, image processing
