Inspiration
When we came in for the hackathon, on Saturday morning, we knew we wanted to do something related to machine learning, so we spent all of Friday night learning different technology stacks and playing with python to do some analysis. We realized that training a data set would take too long if we actually wanted to implement something crazy, so instead, we turned to the machine learning Gods that could help us fix our problems - Google Vision API. On our casual stroll across downtown Atlanta on Saturday morning, we realized that there were so many landmarks that weren't labeled and a quick google search was able to tell us what the landmark was. There had to be a better place to tell us what we were looking at.
What it does
The app uses Google's vision API to detect the content of images, and creates meaning out of them. It
How I built it
We had a 5-step process that we used:
- Upload image to firebase
- Firebase function to connect to Google Vision API
- Retrieve JSON from Google Vision
- Use cloud firestore function to parse .json
- Display image with tags on Android
Challenges I ran into
The part that took us the longest time was parts 2-4, where we had to use Google's vision API to parse the image and json file.
Accomplishments that I'm proud of
We're able to identify things!
What I learned
This was the first time for both team members to use firebase, and also the first time we used the Google Vision API. We had so much fun integrating both of them together!!
What's next for viATL
Next, we would like to create the iOS app instead of just the Android app, we would like users to be able to do a quick google search for the items that they take pictures of.
Log in or sign up for Devpost to join the conversation.