FIEC Learning| Automated Detection of Hazards
Automated Detection of Hazards
In search of typhoons
Summary
For this project we will apply machine learning to convolutional networks, this will execute the task that tropical cyclones will be detected at a considerable distance, in order to be able to warn people who are dedicated to the investigation of these phenomena and to obtain an approximate of how much damage this phenomenon could cause to the places that are near said event.
How We Addressed This Challenge
Our project would be a very helpful tool when using updated meteorological data, allowing the detection and prevention of natural phenomena, specifically tropical cyclones, in turn keeping researchers and the population informed so that they can make decisions that help save lives in extreme cases.
A convolutional network model was developed to detect storms or hurricanes or typhoons or heavy rains on the earth, so that through a mobile application that sends alerts if necessary and is important because it would help people who work at sea or near the coasts. It works by detecting the phenomena already specified and sending alerts through the application to keep people informed. We expect the model to become more and more efficient in training as time goes on.
How We Developed This Project
What inspired your team to choose this challenge?
We chose this topic, because there have been many natural phenomena worldwide that are not recognized in order to be able to prevent major damages and in this way human lives are not lost.
What was your approach to develop this project?
The approach to carry out this project is that we want to inform people so that they do not feel vulnerable, uninformed, and can prevent and avoid being in nearby places when events are detected or predicted.
What tools, coding languages, hardware, software did you use to develop your project?
We use the Adobe XD tool to design the interfaces of how our application would look like, Python as a language, Drawing.io for the diagrams of our approach, Visual studio code to write the code, Google Earth for the visual capture of a part of our country As a reference to the subject of our project, the EARTHDATA by NASA map to visualize and study what a typhoon looks like, finally, hardware we use mid-range laptops which have enough performance to carry out the project.
What problems and achievements did your team have?
The problem we had as a team to carry out this challenge is that we do not know well the behavior of typhoons, it took us a day to investigate how this phenomenon behaved, to study its characteristics and how this phenomenon formed, when we could already understand this, The other problem was getting a good algorithm, so that it fits the problem we’re going to solve. By investigating these phenomena we were able to understand beyond the damages that can cause, how to prevent the greatest possible damage in case of finding ourselves in a risky situation within the range of phenomena, to create an application that can help us be informed at all using meteorological data, many tragedies can be avoided and it would reassure would also many people.
How We Used Space Agency Data in This Project
The data we used as a reference for the development of our proposal were those of EARTHDATA by NASA that provides Global Imaging Services (GIBS), as a reference and guide for programming we use data from a github repository called spaceapps-fenomena_detection, this repository provides a list of resources for the challenge of detecting phenomena, including satellite images, tagged data, tagging tools and sample code for downloading, processing and machine learning images.
Project Demo
Demo of mobile app