Disaster Management using UAV and Deep Learning
We have built a drone with features that help us know the affected area's problems and help people overcome them using Basic Image Processing and Deep learning algorithms over drone technology.
Created on 16th February 2020
•
Disaster Management using UAV and Deep Learning
We have built a drone with features that help us know the affected area's problems and help people overcome them using Basic Image Processing and Deep learning algorithms over drone technology.
The problem Disaster Management using UAV and Deep Learning solves
Our project makes it possible for people stranded during a flood, cyclone or any natural disaster where network connectivity is unavailable, to communicate with the rescuers using RF mode and mesh networking. Most importantly our OpenCV model helps in making heatmaps where intensity is measured in terms of number of people stranded. It helps in efficient rescue operations. These heatmaps also help the people who provide relief during the disaster to efficiently distribute the same instead of randomly distributing it. People can use the SOS feature available on the drone for calling out for help even in case of zero network connectivity.
Challenges I ran into
Our drone is totally programmed by us and we haven't used any pre-programmed drone and hence we had to face a lot of problems as the drone wasn't stable at first. We got over it luckily though by adjusting the propellers and other stuff. Also, the object detection part was too computationally expensive. To overcome this we used a combination of object tracking and object detection which proved to be quite effective.
Technologies used