It's quarantine time, we all stuck in their place don't know what to eat for dinner, but we also had a hard time following the youtube cooking video where we have to keep track of each step and come back to the video for more info. Introducing the Recipe2Go! A fast and convenient way for you to cook your food during this unique time period by getting your fast recipe. Our goal is to let everyone conveniently cook their favorite food without spending too much time checking a specific time frame of a video. Most of the popular cuisine is complicated, so we want to build a system to automate the process of creating a single detailed recipe for everyone so that they can easily follow and enjoy the cooking experience.
What it does
Recipe2Go is a web application that uses computer vision and natural language processing tools to auto-generate a detailed recipe for anyone who wants to quickly start cooking!
How we built it
Challenges we ran into
- We spent a lot of time trying to divide the video into the correct steps with the correlated contents. The core functionalities on how we divide and separate videos into different keyframes were the most challenging part out of all the components.
- How to get the correct procedures and ingredients from the video.
- We spent a lot of time on how we can use the entity sentiment analysis to determine some keywords and then filter out the correct ingredients. We also had a hard time with setting up Redis and Celery because they are new technologies
Accomplishments that we're proud of
The model that we used to find the frames differences within the video was a huge accomplishment. Also, our solution to make sure all the procedures are correct and aligned with the image is also a big one.
What we learned
We learn so much about OpenCV2 as a group and Google Cloud's APIs related to AI and NLP. We also learned about Kubernetes and Celery as a team which is new technologies for everyone.
What's next for Recipe2Go
We definitely need a more optimized model to improve the accuracy of our content. We also need to focus on the improvement of our existing algorithms in object detection and natural language processing. Security can also be improved where we build an additional authentication system for our users. Our product can also be improved by adding more features such as searching for a specific generated file from the past and share the recipe on the social network platform