[6] - Senti-Lytical

• 

  main page

Inspiration

Sentiment analysis using deep learning models has always been something iI've been very fond of!

What it does

Analyzes the sentiment of various household items from 20 different news outlets. With Senti-Lytical, you can either scan the item using the camera on your mobile device(and the model will recognize the object)or by typing it in to find the News' sentiment.

How we built it

Built in Xcode, developed in Swift, and utilized CoreML to train the model.

Challenges we ran into

Dealing with Xcode glitches and VisonKit issues took several hours of debugging.

Accomplishments that we're proud of

Trained the sentiment model in-house with over a 1000 datasets (from Kaggle).

What we learned

To use more efficient algorithms for sending data between view controllers.

What's next for Senti-Lytical

Implement a wider range of objects in the training dataset for more accurate repsonses!

Built With

• coreml
• swift
• visionkit