Teaching the iPhone to See
summary
An iPhone app uses OpenCV cascade classifier to read the Counter Strike radar, training an XML model on roughly 20,000 positive samples and a similar number of negatives; training runs on desktop, the tiny XML runs on-device at frame rate, and the prototype built in three weeks reliably flags enemies after iteration. The workflow includes color filtering (red-channel subtraction), capture of false positives for retraining, and template/edge alternatives rejected for speed or variance.
AI-written summary · about
chapters (12)
- 0:00Computer vision intro
- 4:00Why machine learning excites indie developers
- 9:30Finding the right project for machine learning
- 13:57Choosing vision tools
- 18:33Cascade classifiers
- 21:45Face detection demo
- 26:21Training models
- 31:35Improving detection
- 34:06Training challenges
- 35:09League enemy detection
- 37:14Opportunities and future of machine learning apps
- 39:38Final thoughts & Q&A
related talks
33mComputer Vision, Not Just for Breakfast Anymore
20mTaming the Machine (feat. Nicolas Cage)
38mDeep Learning for detection on a phone
40mDeep Learning on iOS
55mCore ML: Vision
34mThe Potential of IMAGE Processing in Swift for Developers
40mNavigating the Machine Learning Landscape for Apple Devices
12mCascade models in computer vision to boost accuracy and performance