Inspiration
Car Purchase is the 2nd biggest purchase in price in our lives, only behind a home purchase. But, car purchases are more frequent that first-time car-buyer is usually students that are already in high school or that starts college. However, the car information is not very accessible to these younger generation of car buyers. For me, once I was about to start college, my single mom brought me to a local car dealership which only had about 10-20 different models of cars. My mom only had a day off that day. I was forced to pick and buy a car on that day without much knowledge of prices of cars, not knowing the roughly fair price of cars so that I did not even know if I was overpaying or getting a good deal on that car purchase. That's why I made this application so that users can have the car's model, make and year and fair price when they see "that car" that he/she really likes.
What it does
ThatCar.AI is a web application that takes a photo of a car from users and utilize GPT-4-Vision to classify car's make, model and range of years that the car was produced.
How we built it
We built it with Python Flask as back-end and React for front-end. We used GPT-4-Vision API to vehicle classification off of images of vehicles. We used Supabase (Postgres) for user authentication and storage for user credentials.
We used Python ML library, Scikit-Learn, to build a linear regression model to predict car's price based on mileage, model and usageyear (year sold - car's year)
Challenges we ran into
The challenge was to get the sales data of automobiles and streamlining the data. One dataset was examined but it had very low sales price that deviates enormously and the sales price was clearly cut as 500 and 1000. We determined that this was trade-in values when people bought new cars, which affects the sales price very negatively. We threw out this dataset and got another dataset of auto sales.
Accomplishments that we're proud of
We are proud of making a minimum viable product just in time to showcase our features and future plans.
What we learned
We learned how to build a web app using API endpoints that utilize Flask as API endpoints and React as front end.
What's next for ThatCar.AI
The next step for us is to implement user profile including user's age, gender and student status and store data of car look up information when each user scan a car with our app. Then, we will build a solid database of which car is being looked up by which group of people. The business model will be selling that data of car's model, make and year that were looked up by how many people from each group. Also, for the users, with the data, we will show what car is "trending" (looked up more frequently) to users to enhance user experience
Log in or sign up for Devpost to join the conversation.