Inspiration

We wanted to make an app that would teach the average person how to do simple tasks on their car. This would save them money as the person wouldn't have to drive to dealerships, which markup prices on simple jobs.

What it does

The application takes an image of a license plate. Utilizes OCR to convert the image into computer friendly text. The text is then used to get the vin# for the car. From the vin number the app pulls information such as model, make, fuel type etc. This information is key for us to help the user with their car. In the demo the user is struggling with their battery, and the App is able to highlight which component is the battery.

How we built it

We utilized various technologies for our application to work. The application utilizes OCR in order to convert the image of the license plate into computer friendly text, we also used APIs, to get the vin number from the license plate we utilize API, also to get all the information about the car from the vin we use the NHTSA api. Other technologies we used is github pages to host, and languages utilized are Swift, ARKit, HTML, CSS, JS.

Challenges we ran into

One of the first major challenges we ran into was that no one knew AR. We had to learn ARKit and thankfully Apple had great tutorials for ARKit. We tested our app first on water bottles until the application was able to recognize the bottle in 100ms. We also ran into an issue that when we took a picture with the application it wouldn't give us the image file. Instead it gave it to us in 64 bit encoding. This made it extremely difficult to do any OCR on the image.

Accomplishments that we're proud of

We are proud that we learned a lot about AR. It was so much fun learning Apple's ARKit. The first time we tested the image recognition on the water bottle it took 1500ms to recognize and for some angles it wouldn't recognize at all. Eventually we figured out how to optimize the model of the object, so that the app would have the maximum amount of reference points. This lead to a significant improvement on the recognition for the water bottle. All of this allowed us to detect the battery of the car in under 100ms.

What we learned

We learned git, AR and swift.

What's next for cAR

Next step for cAR is expanding support for different services. Such as changing wiper fluid. Also optimizing our image recognition for multiple cars.

Built With

Share this project:

Updates