AutoNation-License Plate Detection Challenge

Detect license tag from any picture and extract customer information

Comment

Inspiration

Recently AI/ML have greatly drawn my attention, so I wanted to build a project that was heavily based in that. Since I am just a beginner in the subject, I wanted to use Google's Vision API to see how it interacted with Python.

What it does

Give an image, it will detect the license plate, extract it and put it in a string, and then look up the customer's information based on that license tag. Lastly, It would display this information, including if the customer had an appointment at that day.

How I built it

It's all built in python, and imports the modules from google.vision in order to use its API. The customer information is in a .json file. There are three main functions in this project: 1: Find License plate, 2: Find Car, & 3: find Customer

Challenges I ran into

-Sometimes Google Vision wouldn't detect the license plate as an object if the car was too far out or if there were other objects in the picture. Solution: I created a function to detect the car as a whole and crop out that part of the picture and ran it again to detect the license plate.

-When trying to extract the text in the license plate, it would sometime include data such as the State, or a sticker that it would have. Solution: Found the text block with the largest letter height, and inferred that as the license tag.

Accomplishments that I'm proud of

I'm happy with both of my solutions for the challenges I encountered, but most importantly I am proud of being able to understand Google's API and how to utilize it to achieve my goals.

What I learned

Surprisingly, the importance of quality photos. Pictures with high resolutions were more easily detectable by Google.

What's next for AutoNation-License Plate Detection Challenge

First, I would like to implement this code using Google Cloud Storage, since I read it can be up tp 50% faster than doing it from a local file. Second, I would like to develop my own AI model to process images and find a particular object/data in it.

Built With

Share this project:

Updates

posted an update —

Good afternoon, I recently received an email that my Google Cloud project was suspended for posting my key in github. It still runs in my computer, but just wanted to post an update in case you run my code and it gives some error.

Log in or sign up for Devpost to join the conversation.