Awards & Nominations
Sweaty has received the following awards and nominations. Way to go!
Automated Detection of Hazards
Fire Watch
Summary
The generalized concept is as follows, the user with some data will create an input file with longitude and latitude with corresponding satellite data of all kinds from wind speed to Chlorophyll-A density, for the initial prototype we use. We attempted to train a neural network on the data from JAXA and CSA but were unable to complete it in time to an acceptable degree of accuracy in the required time frame but ideally, this would replace our statistical method. The algorithm will then populate a world map with red pixels to indicate "hotspots" which are at high risk of wildfires.
How We Addressed This Challenge
With global worming becoming more and more of a problem with every passing day and our team being composed of two South African 20 year old's our future is very dependent on how we deal with this issue. Wild Fires are a massive problem in South Africa (as in much of the world) so we approached the challange from the angle of prevention and prediction of wildfires hoping that we could use satellite data and a ML model to give the average person an easy to interpret and striking representation of the locations which are at risk of Wild fires. In this way we hope that, given any data our algorithm could be expanded to show the influence of many factors such as plant density, humidity and temperature on the size and/or locations of wild fires.
How We Developed This Project
Our project was inspired by a university project where we designed a water simulation over a given terrain. We primariy used Javas' built in libraries for image manipulation and data processing. Our attempted neural network model was built using tensor flow but had poor results given the short time frame and poor computational power for training as well as our team struggling to obtain wildfire data from some of the satelites. We rather used a statistical approch in order to produce a working concept.
How We Used Space Agency Data in This Project
The project relies heavily on data, in fact the entire premise is visualise a variety of complex data so that the lay person can understand the severity of global worming. We have only included Carbon Monoxide readings and temperature measurements in our submission but we hope to expand this to Wind Speed, Rain Fall, Humidity, Chlorophyll-a density and many datasets available from space agency's across the world. We hope that feeding such good quality data to a ML model with high accuracy wild fire data we could produce a model able to highlight non-obvious "hotspots" to better help people with fire prediction, prevention and education.
Project Demo
https://drive.google.com/drive/folders/1nDutU8Ete_STvHtKplmXIdptnz72oC1T?usp=sharing
Project Code
Data & Resources
JAXA and CSA data was used.