Inspiration
Inspiration was the struggling tourism industry that needed something that required to bring it back to life.
What it does
The solution was a tour guiding drone that brings people from place to place, giving facts about the various locations. The drone is able to use transfer learning to recognize the landmarks. It also has a pathfinding algorithm which can lead it from place to place in the best route possible. Lastly, it uses NLP to allow people to search and get answers/facts about the landmarks.
How I built it
It was built entirely on python, as there were many libraries we could tap into.
Challenges I ran into
Initially we wanted to use CNN for the location recognition. But there was insufficient computing power, not enough labelled data, and the trialed CNN that was trained yielded poor prediction results. We decided to use transfer learning instead. Given the time constrain, it was also very difficult to come up with a complex path finding algorithm. We decided to come up with a simple visualization instead.
Accomplishments that I'm proud of
Coming together over 3 days to produce a feasible idea is something we are all proud of. High accuracy of CNN Chat feature works very well.
What I learned
We learnt not to limit ourselves, and not to rule out ideas before we thoroughly think about them.
What's next for Masrine Learning - Drone As a Service
The next step is to build the drone and train it to be self pilot.
Log in or sign up for Devpost to join the conversation.