Inspiration
Our path first started when I asked our team, "We need a great idea, what are some of the things that bother you the most in everyday life?". While GetPic was not conceived at that very moment it is clear that this question was the basis for the rest of our ideas, and especially, GetPic. We knew that our project would have to be something ubiquitous in everyday life as well as something that would apply directly to us college students. Saving money through price matching would benefit us as well as people from all walks of life. From then on, GetPic was born.
What it does
Our app uses computer vision algorithms in conjunction with machine learning optimizations to accurately detect keywords describing the selected object. These keywords are then fed into various e-commerce API's such as Amazon, giving our users their choice of low prices from different vendors.
How we built it
Using Apple's Xcode IDE, we wrote an iOS app in the Swift programming language. We then built an extensible interface that was able to use many different image recognition models, increasing the accuracy of our application. The image recognition models we employed were provided by Clarifai and Google's Cloud Vision API.
Challenges we ran into
The two biggest challenges we encountered were our attempts to use two API's for our item lookups, Amazon's and Walmart's. After working for around 3 hours on implementing Amazon's Product Advertising API (registering as an Amazon Affiliate, working through various hashing problems, and providing a lot of personal information), we weren't able to use that API since it takes 3-4 days for an affiliate to be confirmed so that time was wasted. After this setback, we decided to explore other options, specifically Walmart's Open API. However, after registering for their system many times and requesting multiple confirmation emails, none of which came through, we figured out that none of these most popular API's would work.
Accomplishments that we're proud of
Working with the asynchronous nature of so many image recognition models and compiling all of their results into a single centralized result was a very impressive task.
What we learned
For the more novice members of our group, working on this project allowed them to learn key skills in areas such as VCS principles, different programming languages, and real-world computer science experience. As for the others, we gained a greater understanding of how API's from companies such as Google and Amazon work.
What's next for GetPic
In the future, we plan on incorporating more e-commerce API's, optimizing our detection algorithms with more models (and moving the computation on to a server in the process) as well as integrating payment frameworks to provide a more continuous experience for our users.
Log in or sign up for Devpost to join the conversation.