2023-09-07 22:00:00
Deep learning is a discipline that uses artificial neural networks to automatically learn transformations. Over the past ten years, this discipline has disrupted different areas of data science, such as image analysis or natural language processing, to the point of causing a revival of artificial intelligence (AI). This technological revolution has pushed large companies to recruit researchers and engineers at high prices to form or strengthen AI teams. Many start-ups have also been created to develop solutions to problems that were considered out of reach only a few years ago.
Three-dimensional (3D) images also occupy a growing place in industrial applications, thanks to progress in increasingly efficient acquisition methods, such as X-ray tomography, magnetic resonance imaging or laser remote sensing (more commonly known as LiDAR, from English Light Detection And Ranging). Note that another area of research concerns the extraction of 3D information from 2D images.
It is therefore natural that applications of deep learning for 3D images have been developed in recent years. This article aims to present these methods in a concise and accessible manner. To do this, we start by introducing the different 3D representations considered: we limit ourselves here to representations in the form of a three-dimensional table or graph. We then briefly introduce the basics of deep learning for images and graphs, and then explain how they are applied to 3D images. In the following section, we address a more prospective subject: the generation of 3D images. Finally, before concluding, we discuss the prospects and challenges of these methods.
Use
the reader will find at the end of the article a glossary of important terms and expressions in the article, as well as a table of acronyms, notations and symbols used throughout the article.
1694201559
#Deep #learning #threedimensional #images #Complete #file