Inspiration
We all know someone that has a food allergy or dietary limitation that spends hours at the grocery store looking through large ingredient lists to see if they can consume the food item and limit any possible allergic reactions. Added on top of this hassle, the font sizes are often super small and unreadable for people even with moderate visual issues. We wanted to create something that could help solve both of these problems in a quick and efficient way!
What it does
Our product allows a user to upload a flat, clear image of an ingredient label to our system. Then the user will be prompted to state if they have a dietary need or allergy, such as a sugar-free or vegan diet. After this, our system will use image processing to detect and print the text from the image and state if the food falls under their specific dietary need.
How we built it
We used preexisting image processing libraries such as Open CV, paired with optical character recognition tools such as Pytesseract, to process the text displayed in the image for the first section of the code. Afterwards, we used python libraries and pandas to read files and determine if the user's dietary input was in conflict with any of the ingredients listed in the label image they uploaded.
Challenges we ran into
Since we are all new to ML and image processing, it was hard to find a perfect image processing method and get the features right for every possible image in the time frame that was given. This is why the user needs to input a clear photograph, free of harsh angles or any external obstructions.
Accomplishments that we're proud of
We are so so so proud of our teamwork, staying on task, and focusing on learning something that we have never learned before. It made the experience very worthwhile!!
What we learned
We learned how to use preexisting trained ML models, image processing libraries and python libraries to analyze large data sets and images.
What's next for IngredientAlert!!
If we have more time in the future, we would definitely like to improve the code for the image processing software. We would also like to create an app or some sort of mechanism that makes the program come to life more!
Log in or sign up for Devpost to join the conversation.