Inspiration
In developing countries like India and the Asia Pacific, road navigation have become one of the major problems due to various reasons. Current map providers like Google and Here maps are able to provide only the road congestion information. But there are several other factors which we can consider to improve its health. The health of the road can be a function of space availability, luminous intensity, the density of ditches per square kilometre apart from congestion. Considering all these factors, a holistic system was created which can grade the road based upon these parameters.
What it does
It detects the health of the road by finding out the ditches, speed bumps, potholes and width of the road etc. to find its health and based upon it'll grade the road
How I built it
I collected the data from the roads in Bangalore using Go Pro camera and then fed these video feeds to the neural network which I built
Challenges I ran into
Initially, I tried to run the algorithm on the normal CPU system. But the algorithm didn't converged and then I tried to run it on the GPU machine and it worked. Also labelling the data is also a challenge as it took lots of time to create the training data.
Accomplishments that I'm proud of
What I learned
What's next for Improving Mobility through Smart Navigation
Built With
- deep-learning
- lstm
- python
- rnn
- tensorflow