Inspiration
GreenEats was inspired by the uptick in stressing dieting among teens and people in general. Most of these diets, ironically enough, are unhealthy and unsustainable. We were inspired to create a way for people to maintain their well-being without having to sacrifice their favorite foods so that they can find their healthy balance.
What it does
We've made a website that recommends healthier but more sustainable recipes. Users can log in and input images of their favorite foods, and it’ll return to you with a suggested recipe. You can also access previous recipes that the program has been used for.
How we built it
Our website is built with React. Once the user logs in to the site and selects an image, computer vision and deep learning techniques come into play to identify what food it might be. This algorithm uses TensorFlow. It then compares the identified food to those in a spreadsheet and returns the correct recipe link. To integrate the AI algorithm into the website (and also for the login page), we used Flask.
Challenges we ran into
For some of us, it was our first time using React, so we had little time to familiarize ourselves with it. We also had trouble integrating the front end and back end together.
Accomplishments that we're proud of
Finishing, for one! We're also proud to have developed a machine learning algorithm in TensorFlow that is capable of accurately classifying food in images. By using deep learning techniques, we were able to train the algorithm to recognize a wide range of dishes with high accuracy, even when faced with challenging conditions like lighting variation or partial occlusions.
What we learned
Our biggest learning curve was with React, such as with using it together with Flask.
What's next for GreenEats
In the future, we hope to add a recommendation engine to our website that can suggest healthier alternatives based on previous recipes (instead of recipes solely based on direct user input). We're also planning to implement a feature where people around the world can contribute to the model by uploading pictures of their food and recipes. This will allow the model to recognize an even wider range of ingredients and dishes, making it more powerful and versatile.
Log in or sign up for Devpost to join the conversation.