Cloud Landscape Images Segmentation Using Artificial Neural Networks and Amazon Web Services for Ecological Applications

Cloud computing is an information technology revolution that allows us to access a shared pool of computing resources such as servers, storage and applications over a network providing the necessary infrastructure to power trending technologies such as artificial intelligence, big data and the inter...

Full description

Saved in:
Bibliographic Details
Main Authors: Vidal-Espitia, Ulises, Serrano-Rubio, Juan Pablo, Ruiz, Mateo D.M., Herrera-Guzman, Rafael
Format: Conference Proceeding
Language:English
Subjects:
Cloud computing
Ecosystems
Image segmentation
Linux
Multilayer perceptrons
Real-time systems
Web servers
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cloud computing is an information technology revolution that allows us to access a shared pool of computing resources such as servers, storage and applications over a network providing the necessary infrastructure to power trending technologies such as artificial intelligence, big data and the internet of things. In this paper we present the design of a framework that employs a cloud computing architecture for real-time digital processing of landscape images using artificial intelligence. This framework is the extension of our work integration of remote sensing and image processing to test the effects of disturbances in a mangrove ecosystem after a hurricane [1], using a multilayer perceptron for superpixel classification. Our framework works using the Amazon Web Services platform by creating a Linux instance on Amazon Elastic Compute Cloud (Amazon EC2). A high concurrency web server allows us to store the landscape images to be further processed within the Linux instance using Django, Python libraries and artificial intelligence for classification and counting of superpixels that resemble micropatches of biotic and abiotic components. Our framework is request-based which allows us to perform processing and obtain task results in real time. On average, our framework classifies all superpixels of a landscape image in 1.37 minutes using 13 classes of patterns.
ISSN:2577-1655
DOI:10.1109/SMC52423.2021.9659054