-
Logo
-
Screenshot
Inspiration
For the past week, my parents have been out of town, so I have been eating frozen Costco food and instant noodles for all three meals every day. Although I had raw foods to create meals with, I did not want to enter my ingredients into a website, search through the recipes to find a good meal, find a recipe I actually had all the ingredients for, and most importantly, find a recipe that doesn't take years to create.
What it does
My scenario can be solved using Vision Nutrition! Vision Nutrition is a website that uses your camera for you to take an image of all of your ingredients so you don't have to enter all of your ingredients, identifies your ingredients in the image, and lists all recipes by time and rating(the way the recipes are sorted can be changed). Since convenience is one of our core values, we try to find recipes that take the least amount of time to create and use all of your ingredients yet don't have many more ingredients listed in the recipe. Not only this, it includes a fun nutrition fact every day!
How we built it
Our website is built by using HTML and CSS, and it lists all of the recipes by the sorting method we listed above. The image ingredient identification is done using Tensorflow, where we trained a model to recognize different ingredients through image recognition. The recipes are all scraped from "Allrecipes," and the ingredients recognized are put into their "Include these Ingredients" filter, and then we filter these results by time and rating.
Challenges we ran into
Some challenges that we ran into were that we originally wanted to use google vision for the ingredient identification, however, because of some coding complications, we decided to use Tensorflow instead. Another challenge that we encountered was the recipe scraping we did from Allrecipes.
Accomplishments that we're proud of
The biggest accomplishment that we are proud of is actually creating the product that we had in mind and making it work within the span of two days.
What we learned
We learned how to train a model to not only identify foods but any image with a specific object inside of it. We also learned how to scrape recipes from Allrecipes, and acquire recipes from our queries and sort/filter them.
What's next for Vision Nutrition
In the future, we will improve the UI of Vision Nutrition, the model training for the ingredient identification to increase identification accuracy, and to increase the number of recipes we have(use other websites).
Log in or sign up for Devpost to join the conversation.