AMAZON I.A

https://mandasoares.github.io/Projeto-Nasa/

: https://youtu.be/opROjJtVgTs

We improved an existing NASA Earth data visualization web interface, implementing an AI developed from scratch, capable of differentiating smoke clouds and detecting possible fire outbreaks in the Amazon, using Earth observational data. Due to the scarcity of datasets and time available, we innovated by creating our own datasets, based on the Resources provided by the challenge and the use of Python and YOLOv5 allowing high accuracy of precision in the data. In addition to developing the web interface, implementing interactive graphics for general use and local analysis using hmtl5 and Angular.

All the material analyzed will aim to train software to recognize and monitor possible fires in the Amazon and other forest biomes around the world.

We believe that with the social recognition of the environmental problem of the Amazon and its support, it will be possible to create public measures that aim to minimize forest fires, being capable of implementation in other terrestrial biomes.

TOOLS:

https://angular.io/docs

A.I training:

YOLOv5

Coding languagues:

• Languages: TypeScript (within Angular framework)

Python

Html 5

Existing Earth Data Visualization App used as base:

Example of an Existing Earth Data Visualization App - “How Did it Change”

Deployed versions and open code of it:

http://bit.ly/howdiditchange , or

https://jeronimonunes.github.io/earth-timelapse/

https://neo.sci.gsfc.nasa.gov/

https://worldwind.arc.nasa.gov/

We use as inspiration the image banks and data analysis of the following Space agencies

• Global Imagery Browse Services (GIBS) APIs | Earthdata
• 
  
• NASA NEO API
• 
  
• NASA AppEEARS

Earth Observation Data Management System (Canada)

ESA EO Catalogue

• In particular, INPE resource about the land cover around Brazil. and his interactive map, warning signal, and ways to filter what you want.

Terrabrasilis/INPE

We learned that even with our team not having knowledge about creating websites, it was a new and socially responsible process that we developed from an urgent and emerging demand when it comes to the largest forest in the world. The construction and development of the Amazon I.A. inspired us by the worldwide dissemination by our team at the “Global Media and Information Literacy Week 2022”, which takes place from October 24th to 31st, promoted by UNESCO. We will present our idea in the session “Dialogue with Indigenous Communities: Media and Information Literacy for Trust”. In addition, our project will also be presented to the coordinators of the NASA SERVIR Global-1 Amazon program that has partnerships with the USAID network that are currently working in the Amazon rainforest.

Our approach promotes preventing fires in the Amazon and reducing forest deforestation by warning about fire outbreaks, this is also a political, social and economic problem that not only affects the Brazilian population but also has an impact on the world. We want to show with Amazon I.A. that the consequences of the lack of preservation can be devastating, even causing effects with global reach.

https://colab.research.google.com/drive/1InCw36iUboMbZ20_jWWqPaTq_oMMGXuB

https://summitagro.estadao.com.br/tendencias-e-tecnologia/inteligencia-artificial-preve-queimadas-com-85-de-acerto/

https://github.com/chrieke/awesome-satellite-imagery-dataset

https://www.kaggle.com/datasets/alik05/forest-fire-dataset

https://universe.roboflow.com/roboflow-public/sarnet-search-and-rescue/browse?queryText=split%3Atest+class%3Atarget&pageSize=50&startingIndex=150&browseQuery=true

https://universe.roboflow.com/models/object-detection

 Link do dataset utilizado:

https://complex.ustc.edu.cn/sjwwataset/list.htm

https://github.com/yozorasa/satellite-cloud-detect

#AI #Amazon #citizenscience #Earthobservation #Observationdata