We're college students. After paying tuition, rent, and studying for all our courses, we're left with both little time and little money. That's why we're always on the look out for a deal. We noticed that Google Maps currently integrates ride-sharing services (Lyft, Uber...) when you look up a route. So we thought why not integrate deals for restaurants, stores, and other retail services? We were honestly surprised that Google hadn't already done this! So we did.
What it does
We created an app that augments Google maps with deals. Whenever you search up a restaurant e.g."Popeyes" the coupons and deals are automatically aggregated and displayed! In addition, by leveraging the Twilio API we can send users reminds and links to deals. This offers a convenient tool for shoppers to quickly discover deals.
How we built it
We used React for the frontend, and NodeJS for our backend. Our application is deployed on GCP App Engine (thanks for the free credits Google!) and we use Twilio's API to send reminders (thanks again Twilio, your organizers were extremely helpful and fast). We also planned on using Google's Cloud Vision API to automatically scrap deals from websites (this proved to be too much of a challenge, more on this later).
Challenges we ran into
It's an understatement, but computer vision is hard. We dove head first into computer vision. None of us had experience with computer vision and we were naive in our initial planning, but there's a lesson in every failure. We gained hands on experience with Google's Cloud Vision API, and we are happy that we attempted it.
We learned that there really isn't a central resource for coupon deals, we had to go to each site or app to manually scrap the coupon data.
Accomplishments that we're proud of
We're really proud that we finished our application, and got (almost) everything to work. We ran into some funny errors with our React deployment on App engine, and get request payloads for our backend, which proved to be very interesting to debug and fix.
What we learned
Although we didn't complete our scraper we learned a great deal about computer vision and Google's Cloud Vision API. We also learned about Twilio's extremely easy to use sms messaging API. We also gained some general experience with provisioning and deploying on GCP (we had more experience with AWS but decided to go with Google because of Cloud Vision).
What's next for FoodMaps
Coupon aggregation and parsing was a huge hassle for us. We would likely reach out to stores and partner with them to create a standardized API to scrape deals. Maybe even an affiliate link like what currently Amazon offers so we could get some kickbacks :). Further, we'd like to add analytics on what coupons users used, and even a poll for most wanted discounts, we believe these features would be extremely valuable for retailers.
Currently up on https://constant-wonder-252304.appspot.com as of 9/8/2019 (likely to be down after)