PPR-Parking Pain Reliever

Inspiration

We were astonished by finding that every person in the US waste 50 hours searching for parking. This not only leads to stress but also waste lots of time and money with an inherent impact on climate. This was the main inspiration behind solving the parking pain.

What it does

We made an interface that lets user search for vacant parking spot in various facilities having open and multilevel parking. The solution is best suited but not limited to malls, hospitals, shopping districts, schools and universities. We employ image recognition using neural network from the live feed of the CCTV recordings. The algorithm differentiates between car and vacant position, generates overlap between parking divider and the car. This helps in identifying the particular spot that is then sent to the mobile/web application in real-time. This is also provided with a video to access the parking spot, thus making users life more easy.

How we built it

We have created our frontend using Angular JS providing users with an interface that connects the processed live data and streams and mark vacant parking spots. The data was processed in the backend using Mask-RCNN by matterpot for identifying cars and vacant parking spots.

Challenges we ran into

Getting live training dataset of CCTV.
Were unable to utilize Mask-RCNN module in full efficiency.
Faced hardware problems.
Integration

Accomplishments that we're proud of

We learned to work in a team by distributing responsibilities and at the same time brainstorming about every step with each other. On technical end we learned about Angular JS, FastAPI, Mask-RCNN and Docker. We are proud to deliver a Minimal viable product that is employs cloud services.