An iOS app that uses a machine learning model to predict Emojis based off of your facial expression, all right in the Apple Messages app!
This app is available on the iOS App Store! View App Store preview.
- Display camera preview to the left to show user real-time input.
- Upon app launch, take a picture to use for determining Emojis.
- Display reload button for the user to use an updated photo for processing.
- Does not contain any Storyboards, and instead was written programatically in Swift.
A list of really cool Core ML models can be found at Awesome CoreML Models Github. This is where I first discovered the model that I used, CNNEmotions. This model was developed by Gil Levi and Tal Hassner. More information can be found at https://talhassner.github.io/home/publication/2015_ICMI.
I did consider and try creating my own ML model to go straight from an image to an Emoji classification using Apple's Turi Create Framework. Upon realizing that I might need to label thousands of selfies with Emojis, I decided it might be much more efficient to move on to other projects.
- Continuous learning! Not only would this be super sick, but it would likely also greatly increase accuracy of the Emoji suggestions. However, such a feature would need a backend.
- Highlighted facial features in camera preview. Showing tracking lines in the camera preview would make the app appear smarter.
- Live Emoji suggestions. Instead of having to press the reload button to get new Emoji suggestions, the app would automatically update its suggestions.