Inspiration
2021 was a bad year not just because of COVID, but also there were around 6 major natural disasters in India alone. From the landfall in Gujarat to the floods in Tamil Nadu, India has seen a lot of natural disasters this year. As per the reports of the WEF, in the last 20 years itself, 1.23 million people have lost their lives and countless people have lost their homes, loved ones and livelihoods. This inspired me and when I got to know about this hackathon I decided to do something about this, first, I built my team then in the ideation phase we got some brilliant ideas, and finally we decided to develop a wholesome application that can predict most of the major natural disaster at one place. We built an application that predicts future earthquakes, tsunami, droughts and floods. Once predicted, a notification is immediately sent to the required authorities on mail as well as SMS. Also the predicted locations with the natural disasters are shown in a world map to get a better view on the situations.
What it does
Basically disaster-breaker predicts natural disasters like Earthquakes, Floods, Tsunamis and Droughts before hand and once predicted a notification is sent to the respective authorities as an email as well as an SMS, so that required measures are taken. The predicted natural disasters are shown in a world political map for us to understand better about the world wide situation.
How we built it
Firstly we had to build the machine learning models to make the predictions. In disaster-breaker sensors are used to get the seismic data using sensors like the seismic sensor. Seismic sensors are instruments that measure the motion in the earth when the earth is shaken or the seismic plates move. This data from the seismic sensors is used to train the ML model. Supervised learning was used for this purpose, based on previous seismic data we could predict the future earthquake occurances. Then using live data we keep checking and whenever the model predicts that earthquake might occur, the corresponding authorities are informed. We have trained our ML models on Amazon Sagemaker lab and we have used Twilio for sending thedisaster notifications. We have used cloud to store and get the live data. The hardware modules have been developed using arduino, seismic sensors etc. In case of the flood prediction modules, we mainly consider the rainfall that year as well as the river level of the closest river to them. Tinkercad has been used for the software simulations. We have used randomforest algorithm for earthquake and tsunami prediction and used logistic regression for the drought and flood prediction modules. The web application has been developed using the django framework in python. We have used google cloud as the cloud service provider to store data make predictions.
Challenges we ran into
We have hosted our application for the first time, and we faced a few challenges while doing that but successfully were able to host our application. Also we have used amazon sagemaker lab for the first time, and not a challenge but it took a little time for us to understand how all the configuration is done and what we had to do. Using twilio was also a challenge as we had to make a account for free trial and then use it. Apart from these we had a few challenges in feature engineering etc but we overcame them and passed with flying colours.
Accomplishments that we're proud of
The main accomplishment, we are proud of is that we could build a application that could save lives in real time. We are really proud of it and would be very happy to see it come into production.
What we learned
A lot of things actually, We learnt a lot in this project not just technical stuff, we learnt so many thing like working in a team etc. Apart from these technically also we learnt many things like using the geo-location and making the django application etc And moreover the use of seismic sensors and cloud to do so.
What's next for disaster_breaker
Our application can further be extended to more natural disasters like hurricanes and volcanoes. Centralized servers can be placed at different parts of the world to have our data transfer faster and easier and our application can become location specific geographically. It can also be enhanced by better machine learning algorithms if any are developed and at the rate technology and research in the field is going on right now, there definitely would be more efficient algorithms in the future. So, basically the base for future research has been set, on top of which more and more research can take place. We can also add a chatbot to help common users to post their queries and get answers as required.
Directions to run the application
pip install -r requirements.txt
python manage.py runserver
You can also checkout our hosted application : http://disaster-breaker.herokuapp.com/
Log in or sign up for Devpost to join the conversation.