-
Our Arduino Connection
Inspiration
Once upon a time, a young Matthew Zita tasted the greatest burger he had ever tasted. That burger was the Krabby Patty. Driven by his hunger for more, Matthew and his friends Faris Kazmi, Mohammad Kamal, and Vince Montero decided to use the power of technology to help them attain the Krabby Patty's secret formula from only images. This project was made for all of us who took an Instagram photo of a delicious meal, only to forget what it was called when we craved it most.
What it does and How it does it
The Krabby Patty Secret Formula Finder uses an Arduino to take photos and upload them to a Google Drive server. Then, a seperate script can access that Google Drive and retrieve the image using built in functions in Keras (a library which is built into Tensorflow). Once the image is retrieved, it is passed the MobileNet model for classification. A string with the classification is returned and is used in an automated search for a recipe on foodnetwork.com using Selenium and Beautiful Soup. Once the recipe is received, it would be sent to a web server using post requests, where it is displayed along side the image.
Challenges I ran into
Using Tensorflow proved to be quite challenging, and although the code is relatively simple compared to creating a machine learning model from scratch, many errors and roadblocks got in the way of creating our very own machine learning model. Analyzing the HTML websites for webscraping took a lot of trial and error and attention to detail. Getting the server to accept post requests was also quite challenging.
Accomplishments that I'm proud of
In the end, we more or less achieved our goal which we set out to do. We used all the libraries and platforms which we wished to improve in and learned tons along the way!
What I learned
Along with all the technical knowledge we learned such as webscraping, back end web development, and machine learning basics, we learned the importance of having a solid, detailed plan from the very beginning, we were able to avoid many problems which we faced in the past by paying special attention to these things.
What's next for Krabby Patty Secret Formula Finder
We want to finish building the website, install a better camera on the Arduino, including more search options and details in the webscraper and expanding the search to the entire internet rather than just food network, and creating a machine learning model which is more specialized and accurate in recognizing food.
Log in or sign up for Devpost to join the conversation.