Next Generation| Automated Detection of Hazards
Automated Detection of Hazards
Hurricane identification and classification
Summary
The project consists of integrating the images collected by the satellite, and being able to classify them into the types of hurricanes (hurricane category 1, category 2, etc). The general idea is that when the satellite detects a shape like a hurricane, it sends us the location where it is originating and its category, it can also have the option of obtaining the parameters such as temperature, size, precipitation.When the phenomenon is determined to be a hurricane, send a message or a signal to the website with the latitude, altitude, type of category, of the hurricane. The page can have multiple API connections in order to speed up the communication partand get data in real time.
How We Addressed This Challenge
This intelligent system will change the way a hurricane is detected by the population, giving acces to an accesible platform wich allows to predict an inminent hurricane in real time. The importance of the early detection of natural disasters is something that needs to be recognized. Which led us to think, what could have been different if more projects like this had been in operation during the most devastating hurricanes, would the results have been different? Using two artificial intelligences, it identifies tropical storms and hurricanes on a map and classifies it according to its category, these being moderate, extensive, extreme and catastrophic. We hope to make as our next step to create a web page or application for any device, that can inform and alert about the location and category of the hurricane and be able to keep the community informed.
How We Developed This Project
This project approaches the (CREATE) challenge by creating innovative technologies in response to the need of the population and government organizations to obtain data in real time about natural hazards that may cause human or economic losses.
We had problems in how to identify the geological coordinate by pixel of the image.
Also, in installing the environment to be able to use the appropriate software and how to use the libraries properly.
How We Used Space Agency Data in This Project
The JAXA satellite was used to download the images that were processed and trained with python keras and tensorflow. We downloaded the images by hand and we were able to extract the specific characteristics of each one in an xml.
Project Demo
This is our video where we register progress: Hurricane identification and classification
Data & Resources
We obtained the images from one of the pages provided in the NASA documentation.
And we use the following programs and libraries:
Python 3.7.8, CUDA_GPU, JAXA, Keras, Tensorflow, Virtualenviroment, Numpy.