Inspiration
We were initially inspired by the vast capabilities we could achieve with AWS. We went through several different applications that we could have built and made prototypes of each, however, in the end we felt that this project aligned best with our team's goal of helping people become healthier and more conscious of their nutritional intake.
What it does
Takes an image of nutrition information and: -Tracks your daily caloric intake. -Tracks your major nutritional intake (macro-nutrients) -Give a total daily intake to compare to ideal daily nutritional averages (by the FDA) <---Theoretical
How we built it
- Used AWS with python to see if we could interpret images using ML.
- Looked at the patterns using interpretations and deduced an algorithm to consistently get relevant information from image
- Built a website to both collect and display this information ## Challenges we ran into
- We took way too long to prototype projects/ideas -> forcing us to rush the later processes
- Communication about what exact inputs and outputs needed to be there/when things needed to be done
- Capabilities of team needed to be reevaluated after team member left (Was a mechanical engineer -> needed him to help build initial project which was an automated trash collector)
Accomplishments that we're proud of
First and foremost, we were able to complete the project under a significantly reduced timeline. Secondly, we were able to work with high level machine learning models - especially AWS - to process images and obtain data from them.
What we learned
One of main lessons we've learned is that we must be very quick and decisive with our decisions. It is a competiton where we must think fast due to its nature, and as such we must be able to [...]. Another important take away is that we must have some of our own materials at hand, such as a Raspberry Pi, Arduino set, structural materials for construction, and vice versa. This is to allow us to make sure that we have everything we need to complete our project and avoid the trouble of having to obtain limited to unavaliable resources during the middle of a time crunch. We also learned many new languages and machines such as Arduino, Raspberry Pi, web development, and machine learning.
What's next for rekognition
One important thing is improving the functionality of the website, such as being able to upload and download data quickly and efficiently. Also improving the ability for the machine to recognize some of the materials in the images that we want it to recognize. Lastly we would like a more intuitive way to figure out the location of the item in question within the image.
Log in or sign up for Devpost to join the conversation.