ChatSmart

Online chatting, but better than ever.

Comment

Inspiration

Many people are working or study from home and they communicate with each other through social media platform to ask questions or plan trips in the future. ChatSmart is an app that can help people during the conversation.

What it does

ChatSmart is not defined by any one function. Rather, the many new features that it offers make it unique.

  • Definition checking: Just click on difficult words to check out the wikipedia definition!
  • Sentiment Analysis: All messages are immediately categorized as positive or negative. If you are having a bad day, you can click on the icon beside your message for puppy gifs!
  • Image Recognition: Want to find that meme from a long time ago with a frog? Just search for it in the chat!
  • OCR: Often times, students send supervisors screenshots of code to resolve together. 6 months later, you encounter the same problem but cant find it anymore :( Good thing you can search for the error in the screenshot!
  • Landmark Classification: Remember the trip to the Pyramids but dont want to scroll all the way up to find it, we gotchu.

How we built it

  • Firebase: Used for authentication, Firestore for storing messages and Storage for storing images. Used Google Cloud Platform VM to host backend points.
  • Azure: Used Text Analytics API for sentiment analysis and entity/keyword extraction. Used Computer Vision API for OCR and Object and tag recognition.
  • EdgeML with Rune: Used for serving Landmark Classification model
  • Cockroachlab DB: Used for storing landmark predictions and planned trips
  • Webstack: React, DJango, Rust

Challenges we ran into

The configuration of Rust is time consuming.

Accomplishments that we're proud of

We’re really proud of the amazing designs we created for the app. We’re also really happy that we learned so many new technologies while creating this app.

What we learned

There were many new technologies we learned in the past 36 hours such as Firebase, CockroachDB, Google Cloud Platform, Azure API, EdgeML with Rune.

What's next for ChatSmart

We will add more features to ChatSmart. Allow asynchronous communication for EdgeML with Rune to classify image. Currently, we only use django to execute the rune command on the Google VM instance for new upcoming images one by one. We will connect trip plans on Google calander.

Built With

Share this project:

Updates