Logo

Command Palette

Search for a command to run...

ArchiveFeatured Projects

The Simpsons Tracker: An iOS App for Character Recognition

The Simpsons Tracker: An iOS App for Character Recognition

The Simpsons Tracker: An iOS App for Character Recognition

An iOS app that uses machine learning to recognize and track main Simpsons characters from live streams, videos, or photos

December 25, 2019

The Simpsons Tracker: An iOS App for Character Recognition

Overview

The Simpsons Tracker is an innovative iOS application that utilizes machine learning to identify and track main characters from the beloved animated series The Simpsons. Built with tools like TuriCreate and CreateML, this app can analyze live streams, videos, or photos to detect characters such as Homer, Marge, Bart, Lisa, and Maggie Simpson with impressive accuracy. The project leverages a dataset from Kaggle, The Simpsons Annotated Dataset, to train its recognition model, making it a unique blend of technology and pop culture.

Visual Development

The project includes visual examples that highlight its functionality:

Terminal Output Terminal output displaying character recognition results from a test input, identifying Homer Simpson.

Character Detection A scene with Maggie and Lisa identified, showing confidence scores of 97% and 96%, respectively.

Technical Exploration

The Simpsons Tracker demonstrates several technical accomplishments in machine learning and mobile development:

  • Model Training: The recognition model was trained using TuriCreate, processing annotated images from the Kaggle dataset to learn character-specific features.
  • Real-time Processing: The app excels at processing live streams and videos on the fly, delivering instant character detection.
  • iOS Integration: Using CreateML, the model was optimized and embedded into the iOS app, ensuring efficient performance on mobile devices.

Technical Approach

The Simpsons Tracker was developed using a combination of TuriCreate and CreateML, tailored for machine learning and iOS deployment. The process included:

  • Data Preparation: Curating and annotating the Kaggle dataset to train the model effectively.
  • Model Development: Training the recognition model with TuriCreate to achieve high accuracy.
  • App Implementation: Building the iOS app with Swift and integrating the model via CreateML for seamless real-time processing.

Narrative Concept

The Simpsons Tracker serves as a practical tool and a playful exploration of machine learning’s potential in everyday applications. By focusing on The Simpsons, it transforms a technical achievement into an engaging experience, allowing users to interact with a familiar animated world in a new way. The app highlights how AI can bridge entertainment and technology, offering possibilities for fan engagement or educational use.

Samples

Below are video demonstrations showcasing the app’s real-time character tracking capabilities.