Inspiration

With almost missing my math final last year, we decided that we needed to solve the parking issue. Finding a parking spot is nearly impossible and time consuming. Even self driving cars have to go through all lanes of a parking lot to finally find a spot. With our mobile app, we can help humans and self driving cars find a parking spot with ease.

What it does

Using map data, we were able to detect cars in a parking lot and associate each car with a Google Maps coordinate to show drivers available parking spots.

How we built it

Using our very own machine learning algorithm, we trained a haar-cascade classifier that distinguishes cars in a parking lot from empty spots. Using OpenCV(computer vision) and Python we associated each detected car with a Google Maps coordinate. The coordinates are sent to our Amazon Web Server where the Javascript files overlay red or green rectangles onto our parking lot.

Challenges we ran into

Our machine learning algorithm would crashed several times before finally being trained. Our laptop was not able to handle the large dataset of car images(2000+). After several tries of allocating maximum memory, we were able to detect all cars with a accuracy of 84%.

What's next for ParkIt

Integrate our solution with self driving cars and car companies such as Uber. Self driving cars are able to enter a parking lot but they must search through all rows before finally finding a parking spot wasting time and fuel.

Password for website below

starboy

Challenges

HackPSU 1st, 2nd, 3rd, Amazon AWS Prizes

Share this project:
×

Updates