FindMySpot

• 

  Here the machine learning algorithm reached 93% accuracy

Inspiration

Society wastes too much time driving and searching for parking. We wanted to create a straight forward solution to this problem for J.B. Hunt and by using Google Cloud services.

What it does

The solution analyzes using an aerial view of a parking lot and then calculates the number of open spots remaining. When the user interacts with Dialogflow, it will analyze the inputs and provide them with the best possible parking spot.

How we built it

We implemented a machine learning back end to detect the location of open parking spots and then compared the values with the distances to create an optimal parking suggestion.

Challenges we ran into

Deciding on the best way to communicate with the user. Formatting the machine learning model so that it would be compatible with our training environment. Learning how to train and create logic in dialogue flow. Trying to find suitable API to communicate between the front and back end.

Accomplishments that we're proud of

Increasing the machine learning model so that it could recognize open parking spots up to 93%. Successfully learning and developing a well-trained algorithm to converse with users in Dialogflow.

What we learned

Machine Learning, image analysis, path finding A *, graph traversal, Dialogflow, computer vision, google cloud services

What's next for FindMySpot

Applying the concept to other difficult parking areas in cities and at companies. Eventually it could be applied to autonomous vehicles to park more efficiently.

Built With

• dialogflow
• flask
• python